CN118779652B

CN118779652B - Driving intention prediction model training method, vehicle driving method and device

Info

Publication number: CN118779652B
Application number: CN202410814462.0A
Authority: CN
Inventors: 吴慧颖
Original assignee: Xiaomi Automobile Technology Co Ltd
Current assignee: Xiaomi Automobile Technology Co Ltd
Priority date: 2024-06-21
Filing date: 2024-06-21
Publication date: 2025-07-08
Anticipated expiration: 2044-06-21
Also published as: CN118779652A

Abstract

The present disclosure proposes a training method for a driving intention prediction model, a vehicle driving method and a device, the method comprising: obtaining an initial driving intention prediction model to be trained, and obtaining sample intention data and sample driving data of a sample vehicle to obtain a first training sample of the initial driving intention prediction model; training the initial driving intention prediction model according to the first training sample until the training is completed to obtain a trained candidate driving intention prediction model; obtaining sample driving style data of the sample vehicle, and adjusting the candidate driving intention prediction model according to the sample driving style data to obtain an adjusted target driving intention prediction model. The accuracy and efficiency of driving intention recognition are improved, the performance of the vehicle's driving assistance function is optimized, the driving safety of the vehicle is improved, and the driving experience of the user is optimized.

Description

Training method of driving intention prediction model, vehicle driving method and device

Technical Field

The disclosure relates to the field of data processing, and in particular relates to a training method of a driving intention prediction model, a vehicle driving method and a device.

Background

Along with the development of society, more and more people select vehicles as a travel tool, in a parallel running scene of the vehicles, drivers need to observe road conditions to judge whether the current road conditions can realize parallel running, and in the related technology, the drivers can be assisted in blind area detection of the vehicles through radars and cameras arranged on the vehicles so as to assist reminding the drivers of carrying out parallel running.

Under this scene, camera and radar can discern current existence and merge demand through the driver to turning to the opening of pilot lamp to start the blind area detection function of vehicle, under this scene, when the pilot does not start the pilot lamp, the blind area detection function probably can not start to cause the influence of certain degree to the security that the vehicle was driven.

Disclosure of Invention

The present disclosure aims to solve, at least to some extent, one of the technical problems in the related art.

To this end, a first aspect of the present disclosure proposes a training method of a driving intention prediction model.

A second aspect of the present disclosure proposes a vehicle running method.

A third aspect of the present disclosure proposes a training device of a driving intention prediction model.

A fourth aspect of the present disclosure proposes a vehicle running apparatus.

A fifth aspect of the present disclosure proposes a vehicle.

A sixth aspect of the present disclosure proposes an electronic device.

A seventh aspect of the present disclosure proposes a computer-readable storage medium.

The first aspect of the disclosure provides a training method of a driving intention prediction model, comprising the steps of obtaining an initial driving intention prediction model to be trained, obtaining sample intention data and sample driving data of a sample vehicle to obtain a first training sample of the initial driving intention prediction model, training the initial driving intention prediction model according to the first training sample until training is finished to obtain a trained candidate driving intention prediction model, obtaining sample driving style data of the sample vehicle, and adjusting the candidate driving intention prediction model according to the sample driving style data to obtain an adjusted target driving intention prediction model.

In addition, the training method of the driving intention prediction model provided in the first aspect of the present disclosure may further have the following additional technical features:

According to one embodiment of the disclosure, the method for obtaining sample intention data and sample driving data of a sample vehicle to obtain a first training sample of the initial driving intention prediction model comprises the steps of carrying out parallel intention recognition on the sample intention data to obtain sample parallel intention data and sample non-parallel intention data from the sample intention data, obtaining sample parallel driving data corresponding to the sample parallel intention data and sample non-parallel driving data corresponding to the sample non-parallel intention data from the sample driving data, obtaining a first candidate sample according to the sample parallel intention data and the corresponding sample parallel driving data, obtaining a second candidate sample according to the sample non-parallel intention data and the corresponding sample non-parallel driving data, and obtaining the first candidate sample according to the first candidate sample and the second candidate sample.

According to one embodiment of the disclosure, training the initial driving intention prediction model according to the first training sample until training is finished to obtain a trained candidate driving intention prediction model, wherein the training comprises the steps of inputting the first training sample into the initial driving intention prediction model to obtain a first training result output by the initial driving intention prediction model, obtaining a first sample label of the first training sample to obtain a first training loss of the first training result based on the first sample label, carrying out parameter adjustment on the initial driving intention prediction model according to the first training loss, and returning to obtain a next first training sample to carry out model training on the initial driving intention prediction model after parameter adjustment until training is finished to obtain the trained candidate driving intention prediction model.

According to one embodiment of the disclosure, the method for obtaining the sample driving style data of the sample vehicle and adjusting the candidate driving intention prediction model according to the sample driving style data to obtain an adjusted target driving intention prediction model comprises the steps of obtaining a plurality of preset sample driving styles and initial driving style data of the sample vehicle, carrying out style division on the initial driving style data according to the plurality of sample driving styles to obtain sample driving style data of each of the plurality of sample driving styles, obtaining a candidate driving intention prediction model corresponding to the sample driving style according to any sample driving style, generating a second training sample of the candidate driving intention prediction model according to the sample driving style data under the sample driving style, and training the candidate driving intention prediction model through the second training sample until training is finished to obtain the trained target driving intention prediction model.

According to one embodiment of the disclosure, the method for obtaining a candidate driving intention prediction model corresponding to a sample driving style according to any sample driving style and generating a second training sample of the candidate driving intention prediction model according to sample driving style data under the sample driving style comprises obtaining sample driving style intention data and sample driving style driving data under the sample driving style according to any sample driving style from the sample intention data and the sample driving data, and obtaining the second training sample of the candidate driving intention prediction model according to the sample driving style intention data and the sample driving style driving data.

According to one embodiment of the disclosure, the training of the candidate driving intention prediction model through the second training sample is performed until the training is finished, so that the trained target driving intention prediction model is obtained, and the method comprises the steps of inputting the second training sample into the candidate driving intention prediction model to obtain a second training result output by the candidate driving intention prediction model, obtaining a second sample label of the second training sample to obtain a second training loss of the second training result based on the second sample label, performing parameter adjustment on the candidate driving intention prediction model according to the second training loss, and returning to obtain a next second training sample to perform model training on the candidate driving intention prediction model after parameter adjustment until the training is finished, so that the trained target driving intention prediction model is obtained.

According to one embodiment of the disclosure, the training of the candidate driving intention prediction model according to the second training sample is performed until training is finished, and after the trained target driving intention prediction model is obtained, the method comprises the steps of obtaining a target driving style of a target vehicle according to any sample driving style, responding to the target driving style and the sample driving style matching, and issuing the target driving intention prediction model corresponding to the sample driving style to the target vehicle.

The second aspect of the present disclosure proposes a vehicle driving method, which includes obtaining a target driving intention prediction model configured on a target vehicle, wherein the target driving intention prediction model is obtained based on the training method of the driving intention prediction model proposed in the first aspect, obtaining target driving style data of the target vehicle, and target intention data and target driving data of the vehicle within a target time range, inputting the target intention data, the target driving data and the target driving style data into the target driving intention prediction model, outputting a target predicted driving intention of the target vehicle through the target driving intention prediction model, and performing driving control on the target vehicle based on the target predicted driving intention.

In addition, the vehicle driving method according to the second aspect of the present disclosure may further have the following additional technical features:

According to one embodiment of the disclosure, the method further comprises the steps of identifying whether corresponding update data exists in the target driving intention prediction model, and in response to the fact that the update data exist, updating and adjusting the target driving intention prediction model according to the update data to obtain a new target driving intention prediction model.

According to one embodiment of the disclosure, the driving control of the vehicle based on the target driving intention comprises the step of performing the parallel driving control on the target vehicle through a steering lamp control module and a parallel assistance module of the vehicle in response to the target predicted driving intention to be the parallel driving intention.

According to one embodiment of the disclosure, after the vehicle is controlled to run based on the target driving intention, the method comprises ending the parallel running control of the target vehicle and uploading the target predicted driving intention output by the target driving intention prediction model to a cloud in response to identifying that the parallel running of the target vehicle is ended.

The third aspect of the disclosure provides a training device for a driving intention prediction model, which comprises a first acquisition module, a first training module and a second training module, wherein the first acquisition module is used for acquiring an initial driving intention prediction model to be trained, acquiring sample intention data and sample driving data of a sample vehicle to obtain a first training sample of the initial driving intention prediction model, the first training module is used for training the initial driving intention prediction model according to the first training sample until training is finished to obtain a trained candidate driving intention prediction model, and the second training module is used for acquiring sample driving style data of the sample vehicle, and adjusting the candidate driving intention prediction model according to the sample driving style data to obtain an adjusted target driving intention prediction model.

In addition, the training device for a driving intention prediction model according to the third aspect of the present disclosure may further have the following additional technical features:

According to one embodiment of the disclosure, the first obtaining module is further configured to identify the sample intention data to obtain sample parallel intention data and sample non-parallel intention data from the sample intention data, obtain sample parallel running data corresponding to the sample parallel intention data and sample non-parallel running data corresponding to the sample non-parallel intention data from the sample running data, obtain a first candidate sample according to the sample parallel intention data and the corresponding sample parallel running data, obtain a second candidate sample according to the sample non-parallel intention data and the corresponding sample non-parallel running data, and obtain the first training sample according to the first candidate sample and the second candidate sample.

According to one embodiment of the disclosure, the first training module is further configured to input the first training sample into the initial driving intention prediction model to obtain a first training result output by the initial driving intention prediction model, obtain a first sample tag of the first training sample to obtain a first training loss of the first training result based on the first sample tag, perform parameter adjustment on the initial driving intention prediction model according to the first training loss, and return to obtain a next first training sample to perform model training on the initial driving intention prediction model after parameter adjustment until training is completed, thereby obtaining the trained candidate driving intention prediction model.

According to one embodiment of the disclosure, the second training module is further configured to obtain a plurality of preset sample driving styles and initial driving style data of the sample vehicle, perform style division on the initial driving style data according to the plurality of sample driving styles to obtain respective sample driving style data of the plurality of sample driving styles, obtain a candidate driving intention prediction model corresponding to the sample driving style according to any one of the sample driving styles, generate a second training sample of the candidate driving intention prediction model according to the sample driving style data under the sample driving style, and train the candidate driving intention prediction model through the second training sample until training is completed to obtain the trained target driving intention prediction model.

According to one embodiment of the disclosure, the second training module is further configured to obtain, for any sample driving style, sample driving style intention data and sample driving style driving data under the sample driving style from the sample intention data and the sample driving data, and obtain the second training sample of the candidate driving intention prediction model according to the sample driving style intention data and the sample driving style driving data.

According to one embodiment of the disclosure, the second training module is further configured to input the second training sample into the candidate driving intention prediction model to obtain a second training result output by the candidate driving intention prediction model, obtain a second sample tag of the second training sample to obtain a second training loss of the second training result based on the second sample tag, perform parameter adjustment on the candidate driving intention prediction model according to the second training loss, and return to obtain a next second training sample to continue model training on the candidate driving intention prediction model after parameter adjustment until training is completed, so as to obtain the trained target driving intention prediction model.

According to one embodiment of the disclosure, the system further comprises a issuing module, wherein the issuing module is used for acquiring a target driving style of a target vehicle according to any sample driving style, and issuing a target driving intention prediction model corresponding to the sample driving style to the target vehicle in response to the target driving style being matched with the sample driving style.

The fourth aspect of the present disclosure provides a vehicle driving device, which includes a second obtaining module configured to obtain a target driving intention prediction model configured on a target vehicle, where the target driving intention prediction model is obtained based on the training device of the driving intention prediction model set forth in the third aspect, a third obtaining module configured to obtain target driving style data of the target vehicle, and target intention data and target driving data of the vehicle in a target time range, a prediction module configured to input the target intention data, the target driving data, and the target driving style data into the target driving intention prediction model, and output a target predicted driving intention of the target vehicle through the target driving intention prediction model, and a control module configured to perform driving control on the target vehicle based on the target predicted driving intention.

In addition, the vehicle running device according to the fourth aspect of the present disclosure may further have the following additional technical features:

according to one embodiment of the disclosure, the device further comprises an updating module, wherein the updating module is used for identifying whether corresponding updating data exists in the target driving intention prediction model, and updating and adjusting the target driving intention prediction model according to the updating data to obtain a new target driving intention prediction model in response to the fact that the corresponding updating data exists.

According to one embodiment of the disclosure, the control module is further configured to perform parallel running control on the target vehicle through a turn signal control module and a parallel assist module of the vehicle in response to the target predicting the driving intention as the parallel running intention.

According to one embodiment of the disclosure, the control module is further configured to, in response to identifying that the parallel running of the target vehicle is finished, finish the parallel running control of the target vehicle, and upload the target predicted driving intention output by the target driving intention prediction model to a cloud.

A fifth aspect of the present disclosure proposes a vehicle for implementing the training method of the driving intention prediction model proposed in the first aspect described above and/or the vehicle running method proposed in the second aspect described above.

A sixth aspect of the present disclosure proposes an electronic device comprising a processor, a memory for storing executable instructions of the processor, wherein the processor is configured to execute the instructions to implement the training method of the driving intent prediction model as set forth in the first aspect above and/or the vehicle driving method set forth in the second aspect above.

A seventh aspect of the present disclosure proposes a computer-readable storage medium, which when executed by a processor of an electronic device, enables the electronic device to perform the training method of the driving intention prediction model as set forth in the first aspect above and/or the vehicle running method as set forth in the second aspect above.

According to the training method, the vehicle driving method and the device for the driving intention prediction model, an initial driving intention prediction model to be trained and a first training sample constructed by sample intention data and sample driving data are obtained, the initial driving intention prediction model is subjected to model training through the first training sample to obtain a candidate driving intention prediction model, optionally, sample driving style data of a sample vehicle are obtained, and the candidate driving intention prediction model is adjusted according to the sample driving style data to obtain a target driving intention prediction model. According to the method and the device, the target driving intention prediction model is obtained through training of the initial driving intention prediction model and adjustment of the candidate driving intention prediction model, so that the target driving intention prediction model can learn the relation between the behavior data of the driver and the driving intention, and therefore the driving intention of the driver is predicted.

It should be understood that the description herein is not intended to identify key or critical features of the embodiments of the present disclosure, nor is it intended to be used to limit the scope of the present disclosure. Other features of the present disclosure will become apparent from the following specification.

Drawings

The foregoing and/or additional aspects and advantages of the present disclosure will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings, in which:

FIG. 1 is a flow chart of a training method of a driving intent prediction model according to an embodiment of the disclosure;

FIG. 2 is a flow chart of a training method of a driving intent prediction model according to another embodiment of the present disclosure;

FIG. 3 is a flow chart of a method of driving a vehicle according to an embodiment of the disclosure;

FIG. 4 is a flow chart of a method of driving a vehicle according to an embodiment of the disclosure;

FIG. 5 is a flow chart of a method of driving a vehicle according to another embodiment of the present disclosure;

FIG. 6 is a schematic structural diagram of a training device for driving intent prediction model according to an embodiment of the present disclosure;

fig. 7 is a schematic structural view of a vehicle running apparatus according to an embodiment of the present disclosure;

Fig. 8 is a block diagram of an electronic device of an embodiment of the present disclosure.

Detailed Description

Embodiments of the present disclosure are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and intended for the purpose of explaining the present disclosure and are not to be construed as limiting the present disclosure.

The following describes a training method of a driving intention prediction model, a vehicle driving method and a device according to embodiments of the present disclosure with reference to the accompanying drawings.

Fig. 1 is a flow chart of a training method of a driving intention prediction model according to an embodiment of the disclosure, as shown in fig. 1, the method includes:

s101, acquiring an initial driving intention prediction model to be trained, and acquiring sample intention data and sample driving data of a sample vehicle to obtain a first training sample of the initial driving intention prediction model.

In the embodiment of the disclosure, the driving intention of the driver can be identified through the related information of the driver during driving of the vehicle, and in the scene, the vehicle can realize auxiliary driving of the driver based on the auxiliary function of the vehicle after the driving intention of the driver is identified.

Alternatively, a model provided with driving intention recognition and driving assistance functions and configured on the vehicle may be marked as a driving intention prediction model on the vehicle, wherein if the model has a need for training, the driving intention prediction model to be trained may be marked as an initial driving intention prediction model.

In the embodiment of the disclosure, a sample for training the initial driving intention prediction model may be marked as a first training sample, and a vehicle for acquiring the first training sample may be marked as a sample vehicle.

Alternatively, the corresponding intention data and running data may be obtained from the sample vehicle and marked as sample intention data and sample running data of the sample vehicle, respectively, and the sample intention data and the sample running data may be processed based on a construction method of the training sample in the related art, so as to obtain a first training sample constructed by the sample intention data and the sample running data.

S102, training the initial driving intention prediction model according to the first training sample until training is finished to obtain a trained candidate driving intention prediction model.

In the embodiment of the disclosure, according to the model training method in the related art, model training is performed on the initial driving intention prediction model through the first training sample until training is finished, and the trained model is determined to be a candidate driving intention prediction model.

Optionally, the training ending condition of the initial driving intention prediction model may be set based on the training round, training is performed on the model of the current round, if the training round matches with the preset training ending condition, training on the initial driving intention prediction model may be ended, and the model obtained by the last round training ending is determined as the trained candidate driving intention prediction model.

Optionally, a corresponding training ending condition may be set according to a result of model training output, and for model training of a current round, if an output result of the round model is matched with a preset training ending condition, training of the initial driving intention prediction model may be ended, and a model obtained by ending the last round training is determined as a trained candidate driving intention prediction model.

S103, sample driving style data of the sample vehicle are obtained, and the candidate driving intention prediction model is adjusted according to the sample driving style data, so that an adjusted target driving intention prediction model is obtained.

In the embodiment of the disclosure, different driving styles may exist among different drivers, wherein the driving styles of the drivers are related to information such as vehicle speed, stepping force of brake, steering angle of the vehicle and the like when the vehicle is driven.

As one example, the vehicle driving style of the set driver may include a firm style, a aggressive style, and a comfortable style, and it is understood that in a scene where the vehicle needs to travel in parallel, the firm style driver generally travels in parallel at a smaller steering angle and a lower vehicle speed when the vehicle travels in parallel.

And, the driver in the aggressive style usually performs the parallel running at a large steering angle and a high vehicle speed when performing the parallel running, and the driver in the comfortable style pays attention to the energy consumption when performing the vehicle driving, and the vehicle speed and the steering angle adopted in the parallel running are between the conservative style and the aggressive style.

Alternatively, the relevant driving data of the sample vehicle in the sample time range may be obtained and analyzed in the driving style dimension, so as to obtain sample driving style data of the sample vehicle, where the sample driving style data may be determined by vehicle speed data, steering angle data, and the like in the relevant driving data of the sample vehicle.

In this scenario, the candidate driving intention prediction model may be adjusted according to the sample driving style data of the sample vehicle, so that the candidate driving intention prediction model may learn the relationship between the driving intention and the driving style.

Further, a model obtained after adjustment according to the sample driving style data is determined as an adjusted target driving intention prediction model.

The method comprises the steps of obtaining an initial driving intention prediction model to be trained, a first training sample constructed by sample intention data and sample driving data, carrying out model training on the initial driving intention prediction model through the first training sample to obtain a candidate driving intention prediction model, optionally obtaining sample driving style data of a sample vehicle, and adjusting the candidate driving intention prediction model according to the sample driving style data to obtain a target driving intention prediction model. According to the method and the device, the target driving intention prediction model is obtained through training of the initial driving intention prediction model and adjustment of the candidate driving intention prediction model, so that the target driving intention prediction model can learn the relation between the behavior data of the driver and the driving intention, and therefore the driving intention of the driver is predicted.

In the above embodiment, regarding the acquisition of the target driving intention prediction model, it may be further understood with reference to fig. 2, and fig. 2 is a flowchart of a training method of driving intention prediction model according to another embodiment of the disclosure, as shown in fig. 2, and the method includes:

S201, sample intention data and sample running data of a sample vehicle are acquired to obtain a first training sample of an initial driving intention prediction model.

Optionally, the sample intention data is subjected to parallel intention recognition to obtain sample parallel intention data and sample non-parallel intention data from the sample intention data.

In the embodiment of the disclosure, the sample intention data of the sample vehicle may include a parallel intention and a non-parallel intention, wherein the parallel intention may be marked as sample parallel intention data in the sample intention data and the non-parallel intention may be marked as sample non-parallel intention data in the sample intention data.

The data acquisition device configured on the vehicle may be used to acquire sample intention data of a driver of the sample vehicle, for example, eye movement direction data of the driver may be acquired by an eye movement instrument configured in a cabin of the vehicle to obtain sample intention data of the driver, and the camera configured on the vehicle may be used to acquire head torsion angle data of the driver to obtain sample intention data of the driver, which is not limited herein.

Optionally, from the sample travel data, sample parallel travel data corresponding to the sample parallel intention data and sample non-parallel travel data corresponding to the sample non-parallel intention data are acquired.

In the embodiment of the disclosure, the sample driving data of the sample vehicle may include driving data when the sample vehicle performs parallel driving, and may also include driving data when the sample vehicle performs non-parallel driving, in this scenario, the sample driving data may be divided according to sample parallel intention data and sample non-parallel intention data identified in the sample intention data, so as to obtain corresponding data of the sample parallel intention data in the sample driving data and mark the corresponding data as sample parallel driving data, and obtain corresponding data of the sample non-parallel intention data in the sample driving data and mark the corresponding data as sample non-parallel driving data.

The method comprises the steps of acquiring a sampling time range of sampling parallel intention data on a sample vehicle, acquiring running data based on the sampling time range from sample running data of the sample vehicle, and determining the part of the data as sample parallel running data corresponding to the sampling parallel intention data.

And acquiring an acquisition time range of the sample non-parallel intention data acquired on the sample vehicle, acquiring running data acquired in the same acquisition time range from sample running data of the sample vehicle, and determining the partial data as sample non-parallel running data corresponding to the sample non-parallel intention data.

Accordingly, the vehicle coordinate when the sample parallel intention data is acquired on the sample vehicle can be acquired, the running data acquired on the same vehicle coordinate is acquired from the sample running data of the sample vehicle, and the part of the data is determined to be the sample parallel running data corresponding to the sample parallel intention data.

And acquiring vehicle coordinates of the sample non-parallel line intention data acquired on the sample vehicle, acquiring running data acquired on the same vehicle coordinates from sample running data of the sample vehicle, and determining the part of the data as sample non-parallel line running data corresponding to the sample non-parallel line intention data.

The sample driving data may include driving habit data of a driver of the sample vehicle, such as steering wheel angle, steering wheel rotation speed, steering wheel grip, brake pedal stroke, brake pedal opening change rate, accelerator pedal stroke, accelerator pedal opening change rate, and steering lamp lever state data, and vehicle body dynamic driving data of the sample vehicle, such as vehicle speed, longitudinal acceleration, lateral acceleration, and yaw rate data, and may include other data that may describe a driving state of the vehicle, which is not specifically limited herein.

Optionally, according to the sample merging intention data and the corresponding sample merging running data, a first candidate sample is obtained, and according to the sample non-merging intention data and the corresponding sample non-merging running data, a second candidate sample is obtained.

According to the embodiment of the disclosure, a training positive sample of an initial driving intention prediction model can be constructed according to sample parallel intention data and corresponding sample parallel running data, and the training positive sample is marked as a first candidate sample.

And constructing a training negative sample of the initial driving intention prediction model according to the sample non-parallel intention data and the corresponding sample non-parallel running data, and marking the training negative sample as a second candidate sample.

The sample parallel intention data and the corresponding sample parallel running data can be processed according to a sample construction method in the related art, so that a first candidate sample constructed by the sample parallel intention data and the corresponding sample parallel running data is obtained, and the sample non-parallel intention data and the corresponding sample non-parallel running data are processed according to a sample construction method in the related art, so that a second candidate sample constructed by the sample parallel intention data and the corresponding sample parallel running data is obtained.

Optionally, a first training sample is obtained from the first candidate sample and the second candidate sample.

In the embodiment of the disclosure, there is a corresponding training sample set for training of the initial driving intention prediction model, and samples in the training sample set may be marked as first training samples.

In the scene, a sample set consisting of a first candidate sample and a second candidate sample can be obtained, and the sample set is marked as a training sample set of an initial driving intention prediction model, so that a first training sample in the training sample set is obtained.

S202, training the initial driving intention prediction model according to the first training sample until training is finished to obtain a trained candidate driving intention prediction model.

Optionally, the first training sample is input into the initial driving intention prediction model to obtain a first training result output by the initial driving intention prediction model.

In the embodiment of the disclosure, the first training sample may be input into an initial driving intention prediction model, the feature extraction is performed on the first training sample through the initial driving intention prediction model, and the intention prediction is performed according to the extracted feature.

Further, an output result of the initial driving intention prediction model obtained based on the first training sample is marked as a first training result.

Optionally, a first sample tag of the first training sample is obtained to obtain a first training result based on a first training loss of the first sample tag.

In the embodiment of the disclosure, the label information of the first training sample may be acquired and marked as the first sample label of the first training sample.

The first sample tag may be used to indicate whether the driving data in the first training sample is driving data under the parallel line intention.

In a scenario where the first training sample is a sample constructed of sample parallel line intention data and sample parallel line travel data, tag information obtained based on the sample parallel line intention data may be used as a first sample tag of the first training sample in the scenario.

Accordingly, in a scenario where the first training sample is a sample constructed of sample non-parallel intent data and sample non-parallel travel data, tag information obtained based on the sample non-parallel intent data may be used as a first sample tag of the first training sample in the scenario.

In the embodiment of the disclosure, the first training result and the first sample tag may be subjected to algorithm processing according to a loss value acquisition algorithm in the related art, so that a loss value of the first training result based on the first sample tag is obtained according to the result of the algorithm processing, and the loss value is determined as the first training loss.

Optionally, parameter adjustment is performed on the initial driving intention prediction model according to the first training loss, the next first training sample is returned to be obtained to continuously perform model training on the initial driving intention prediction model after parameter adjustment until training is finished, and a trained candidate driving intention prediction model is obtained.

In the embodiment of the disclosure, the model parameters of the initial driving intention prediction model can be adjusted according to the first training loss, and the initial driving intention prediction model after parameter adjustment is returned to obtain the next training sample to continue model training until the training is finished.

The relevant content of the end of training the initial driving intent prediction model may be understood in combination with the relevant content in the above embodiment, and will not be described herein.

Further, a model obtained after the initial driving intention prediction model training is completed is determined as a trained candidate driving intention prediction model.

S203, sample driving style data of the sample vehicle are obtained, and the candidate driving intention prediction model is adjusted according to the sample driving style data, so that an adjusted target driving intention prediction model is obtained.

Optionally, acquiring a plurality of preset sample driving styles and initial driving style data of the sample vehicle, and performing style division on the initial driving style data according to the plurality of sample driving styles to obtain respective sample driving style data of the plurality of sample driving styles.

In the embodiment of the disclosure, the driving styles of different drivers are different, wherein driving style analysis can be performed through the data set of the open source, so that a plurality of possible driving styles of the drivers are obtained and marked as a plurality of sample driving styles.

In this scenario, the driving style data of the sample vehicle may be collected, and the collected driving style data may be classified according to a plurality of sample driving styles, where the driving style data collected on the sample vehicle may be marked as initial driving style data of the sample vehicle.

Further, the initial driving style data are clustered according to the plurality of sample driving styles, driving style data in the clusters can be obtained for any one of the clusters obtained by clustering, and the driving style data are marked as sample driving style data under the sample driving style corresponding to the clusters.

It should be noted that, the sample vehicle may be a plurality of preset test vehicles, in this scenario, initial driving style data uploaded by each of the plurality of test vehicles may be continuously received within a set time range, and after the partial data is received, the partial data is classified according to a plurality of preset sample driving styles, so as to obtain sample driving style data of each of the plurality of sample driving styles.

Optionally, for any sample driving style, a candidate driving intention prediction model corresponding to the sample driving style is obtained, and a second training sample of the candidate driving intention prediction model is generated according to the sample driving style data under the sample driving style.

In the embodiment of the disclosure, after the initial driving intention prediction model is trained to obtain the candidate driving intention prediction model, driving style classification can be performed on the candidate driving intention prediction model, which can be understood that after a plurality of initial driving intention prediction models are obtained, model training is performed on the plurality of initial driving intention prediction models respectively to obtain a plurality of candidate driving intention prediction models, and further, driving style dimension classification is performed on the plurality of candidate driving intention prediction models according to a plurality of sample driving styles to obtain candidate driving intention prediction models corresponding to the plurality of sample driving styles.

In this scenario, a training sample of the candidate driving intent prediction model may be constructed from the sample driving style data and labeled as a second training sample.

Optionally, for any sample driving style, sample driving style intention data and sample driving style travel data under the sample driving style are acquired from the sample intention data and the sample travel data.

In the embodiment of the disclosure, the sample intention data and the sample driving data include intention data and driving data under a plurality of sample driving styles, and in the scene, data under the same style as the sample driving style intention data and the sample driving style driving data under the sample driving style can be obtained from the sample intention data and the sample driving data acquired from a sample vehicle aiming at any sample driving style.

Optionally, a second training sample of the candidate driving intention prediction model is obtained according to the sample driving style intention data and the sample driving style driving data.

In the embodiment of the disclosure, the sample driving style intention data and the sample driving style driving data can be subjected to algorithm processing according to a sample construction algorithm in the related art, and further, samples constructed by the sample driving style intention data and the sample driving style driving data are obtained according to the result of the algorithm processing and are used as second training samples of the candidate driving intention prediction model.

Optionally, training the candidate driving intention prediction model through the second training sample until the training is finished, and obtaining a trained target driving intention prediction model.

The second training sample can be input into the candidate driving intention prediction model to obtain a second training result output by the candidate driving intention prediction model.

In the embodiment of the disclosure, the feature information of the second training sample may be extracted through the candidate driving intention prediction model, the intention prediction is performed according to the extracted feature, and the predicted result is used as the output result of the candidate driving intention prediction model, where the output result may be determined as the second training result output by the candidate driving intention prediction model.

Optionally, obtaining a second sample label of the second training sample to obtain a second training result, performing parameter adjustment on the candidate driving intention prediction model according to the second training loss based on the second training loss, and returning to obtain a next second training sample to continue model training on the candidate driving intention prediction model after parameter adjustment until training is finished, so as to obtain a trained target driving intention prediction model.

In the embodiment of the disclosure, the label information of the second training sample may be determined as a second sample label of the second training sample.

Further, algorithm processing of the loss value is performed on the second sample tag and the second training result according to a loss value acquisition algorithm in the related art, and then the loss value of the second training result based on the second sample tag is obtained as a second training loss according to the result of the algorithm processing.

Optionally, the model parameters of the candidate driving intention prediction model may be adjusted according to the second training loss, and the next second training sample is obtained to continue training the candidate driving intention prediction model after parameter adjustment until the training is finished.

The training end condition of the candidate driving intention prediction model may be set according to the training round, and the training end condition of the candidate driving intention prediction model may be set according to the training output result, which is not particularly limited herein.

Optionally, for the current training round, if the candidate driving intention prediction model of the round meets a preset training ending condition, model training on the candidate driving intention prediction model can be ended, and a model obtained after the current round training is ended is determined to be a trained target driving intention prediction model.

After training to obtain the target driving intention prediction model of each of the plurality of sample driving styles, the vehicle to be modeled may be modeled according to the driving style of the vehicle.

Optionally, for any sample driving style, a target driving style of the target vehicle is obtained.

In this scenario, historical driving style data of the target vehicle may be acquired and analyzed, so that the driving style of the target vehicle is obtained according to the analysis result and marked as the target driving style.

Optionally, in response to the target driving style matching the sample driving style, issuing a target driving intention prediction model corresponding to the sample driving style to the target vehicle.

In the embodiment of the disclosure, for any sample driving style, when the target driving style of the target vehicle is the same as or similar to the sample driving style, it may be determined that the target driving style of the target vehicle matches the sample driving style.

In this scenario, a target driving intention prediction model in the sample driving style may be acquired, and the target driving intention prediction model may be issued and configured to the target vehicle.

The training method of the driving intention prediction model is provided, so that the target driving intention prediction model can learn the relation between the driving intention and the driving intention, thereby realizing the prediction of the driving intention of the driver, improving the accuracy and the efficiency of driving intention recognition, reducing the possibility of potential safety hazards in the driving of the vehicle caused by the fact that the driving intention of the vehicle is not recognized, optimizing the performance of the driving auxiliary function of the vehicle, improving the driving safety of the vehicle and optimizing the driving experience of a user compared with the method of recognizing the driving intention only through the starting of the steering lamp of the vehicle in the related art.

The embodiment of the present disclosure further provides a vehicle driving method, which can be understood with reference to fig. 3, and fig. 3 is a schematic flow chart of the vehicle driving method according to an embodiment of the present disclosure, as shown in fig. 3, and the method includes:

s301, a target driving intention prediction model configured on a target vehicle is obtained.

The target driving intention prediction model is obtained based on the training method of the driving intention prediction model proposed by the embodiment of fig. 1 to 2.

In the embodiment of the present disclosure, a vehicle configured with a target driving intention prediction model may be marked as a target vehicle, in which case the target vehicle may predict the driving intention of the driver through the target driving intention prediction model configured thereof.

S302, target driving style data of a target vehicle, target intention data and target driving data of the target vehicle in a target time range are acquired.

In the embodiment of the disclosure, the historical driving data of the target vehicle can be obtained, the driving style of the target vehicle is obtained through analysis of the historical driving data, and the driving style is determined as the target driving style of the target vehicle.

In this scenario, the relevant data of the target vehicle in the target driving style may be acquired, and the partial data may be marked as target driving style data of the target vehicle.

Alternatively, a time range in which the target vehicle needs driving intention recognition prediction may be marked as a target time range, and in this scenario, intention data and travel data of the target vehicle in the target time range may be acquired and respectively marked as target intention data and target travel data in the target time range.

The target intention data may include at least eye movement data, head torsion angle data, and the like of a driver of the target vehicle, and the target travel data may include at least steering wheel angle, steering wheel speed, steering wheel grip, brake pedal stroke, brake pedal opening change rate, accelerator pedal stroke, accelerator pedal opening change rate, and turn signal lever state data of the target vehicle, and speed, longitudinal acceleration, lateral acceleration, yaw rate, and the like of the target vehicle, which are not particularly limited herein.

S303, inputting the target intention data, the target running data and the target driving style into a target driving intention prediction model, and outputting the target predicted driving intention of the target vehicle through the target driving intention prediction model.

In the embodiment of the present disclosure, the target driving intention prediction model is a model for performing driving intention prediction of a driver, which is configured on a target vehicle, in which the acquired target intention data, target travel data, and target driving style data may be input into the target driving intention prediction model configured on the target vehicle.

The present driving intention of the driver of the target vehicle is predicted based on the target intention data, the target travel data, and the target driving style data by the model capability of the target driving intention prediction model, and the output prediction result is determined as the target predicted driving intention of the target vehicle.

The target predicted driving intention may be a concurrent driving intention of the target vehicle, or may be a concurrent driving intention of the target vehicle.

S304, running control is performed on the target vehicle based on the target predicted driving intention.

In the embodiment of the disclosure, the parallel auxiliary driving module is arranged on the target vehicle, and in the scene, after the target driving intention prediction model outputs the target predicted driving intention, the parallel auxiliary driving module can perform driving control on the target vehicle according to the output target predicted driving intention so as to realize auxiliary driving of a driver.

Optionally, the hardware configuration of the target vehicle may be adjusted based on a preset running control policy to implement running control of the target vehicle, or the central control software system configured on the target vehicle may be adjusted, so that running control of the target vehicle is implemented through the central control software system of the target vehicle, which is not specifically limited herein.

According to the vehicle driving method, the target driving intention prediction model configured on the target vehicle is obtained, the target driving style data of the target vehicle and the target intention data and the target driving data of the target vehicle in the target time range are obtained, the target predicted driving intention of the target vehicle is obtained through the target driving intention prediction model according to the target style data, the target intention data and the target driving data, and then the driving control is carried out on the target vehicle according to the target predicted driving intention. In the present disclosure, the driving intention prediction model of the target vehicle obtained by the method of the embodiments of fig. 1 to 2 predicts the driving intention of the target vehicle, which improves the accuracy and efficiency of the driving intention prediction of the target vehicle, and in the case that the auxiliary driving function of the vehicle cannot be started due to the fact that the driving intention of the driver cannot be recognized is avoided in the case that the auxiliary driving function of the vehicle is started by recognizing the driving intention of the vehicle, improves the stability of the driving intention prediction function and the auxiliary driving function of the vehicle, and optimizes the driving experience of the user.

In the above-described embodiment, regarding the driving control of the target vehicle, it may be further understood with reference to fig. 4, and fig. 4 is a schematic flow chart of a driving method of a vehicle according to another embodiment of the disclosure, as shown in fig. 4, and the method includes:

s401, in response to the target predicted driving intention as the parallel driving intention, the parallel driving control is performed on the target vehicle through a steering lamp control module and a parallel auxiliary driving module of the vehicle.

In the embodiment of the disclosure, the target predicted driving intention of the target vehicle may be a parallel driving intention, in which, when the target predicted driving intention of the target vehicle is recognized as the parallel driving intention, driving control of the target vehicle may be achieved through control of related hardware devices such as a turn signal of the target vehicle, and driving control of the target vehicle in the scene may be marked as parallel driving control.

As an example, as shown in fig. 5, the end target driving intention prediction model collects target intention data, target travel data, and target driving style data of the target vehicle through the data processing module shown in fig. 5, and transmits the collected data to the end target driving intention prediction model shown in fig. 5, predicts the driving intention of the target vehicle through the end target driving intention prediction model shown in fig. 5, thereby outputting a target predicted driving intention.

In this example, when the target predicts that the driving intention is the parallel running intention, the turn signal of the target vehicle may be turned on by the turn signal control module shown in fig. 5, and the auxiliary running of the parallel running of the target vehicle may be performed by the parallel auxiliary running module shown in fig. 5.

The parallel auxiliary driving module can detect a driving blind area of the target vehicle after recognizing that the target vehicle has the parallel driving intention, so as to improve the driving safety of the target vehicle in the parallel driving process.

And S402, ending the parallel running control of the target vehicle in response to the recognition of the end of the parallel running of the target vehicle, and uploading the target predicted driving intention output by the target driving intention prediction model to the cloud.

In the embodiment of the disclosure, when the parallel running of the target vehicle is finished, the parallel running control of the target vehicle may be finished by the turn signal control module and the parallel auxiliary running module shown in fig. 5.

The turn signal control module shown in fig. 5 may turn off the turn signal that the target vehicle has turned on, and stop running the auxiliary running function of the parallel auxiliary running module shown in fig. 5, so as to end the parallel running control of the target vehicle.

Optionally, after the target vehicle finishes the parallel running, the target predicted driving intention output by the vehicle-end target driving intention prediction model may be uploaded to the cloud end through a data transceiver module configured on the target vehicle, where the data transceiver module may be an edge computing module configured on the target vehicle, or may be another functional module configured on the target vehicle and capable of performing data transceiver with the cloud server, and the specific limitation is not herein.

It should be noted that, there is a need for updating the target driving intention prediction model configured on the target vehicle, where whether the target driving intention prediction model has corresponding update data may be identified, and in response to identifying that the update data exists, the target driving intention prediction model is updated and adjusted according to the update data, so as to obtain a new target driving intention prediction model.

In the embodiment of the disclosure, data generated by the cloud for updating the target driving intention prediction model configured at the vehicle end may be determined as update data corresponding to the target driving intention prediction model.

As an example, as shown in fig. 5, setting the update optimization of the model by the cloud target driving intention prediction model shown in fig. 5, in this example, the vehicle end target driving intention prediction model needs to follow the cloud target driving intention prediction model for update optimization due to the update optimization of the cloud target driving intention prediction model.

In this example, the optimization data of the cloud end target driving intention prediction model based on the vehicle end target driving intention prediction model may be determined as update data of the vehicle end target driving intention prediction model, and the update data may be issued to the vehicle end target driving intention prediction model through a data transceiving link therebetween.

Further, the vehicle-end target driving intention prediction model performs updating optimization of the model based on the received updating data, and the updated and optimized model is determined to be a new target driving intention prediction model of the vehicle end.

According to the vehicle driving method, when the target predicted driving intention is recognized as the parallel driving intention, the parallel driving control is carried out on the target vehicle, so that the situation that the auxiliary driving function of the vehicle cannot be started due to the fact that the driving intention of the driver cannot be recognized is avoided, the stability of the driving intention predicting function and the auxiliary driving function of the vehicle is improved, and the driving experience of a user is optimized.

The training method of the driving intent prediction model according to the above embodiments corresponds, and one embodiment of the present disclosure further provides a training device of the driving intent prediction model according to the embodiment of the present disclosure, and since the training device of the driving intent prediction model according to the embodiment of the present disclosure corresponds to the training method of the driving intent prediction model according to the above embodiments, implementation of the training method of the driving intent prediction model is also applicable to the training device of the driving intent prediction model according to the embodiment of the present disclosure, and will not be described in detail in the following embodiments.

Fig. 6 is a schematic structural diagram of a training device for a driving intent prediction model according to an embodiment of the present disclosure, as shown in fig. 6, a training device 600 for a driving intent prediction model includes a first obtaining module 61, a first training module 62, and a second training module 63, where:

A first obtaining module 61, configured to obtain an initial driving intention prediction model to be trained, and obtain sample intention data and sample driving data of a sample vehicle, so as to obtain a first training sample of the initial driving intention prediction model;

The first training module 62 is configured to train the initial driving intention prediction model according to the first training sample until training is completed to obtain a trained candidate driving intention prediction model;

The second training module 63 is configured to obtain sample driving style data of the sample vehicle, and adjust the candidate driving intention prediction model according to the sample driving style data, so as to obtain an adjusted target driving intention prediction model.

In the embodiment of the disclosure, the first obtaining module 61 is further configured to identify a merging intention of the sample intention data to obtain sample merging intention data and sample non-merging intention data from the sample intention data, obtain sample merging running data corresponding to the sample merging intention data and sample non-merging running data corresponding to the sample non-merging intention data from the sample running data, obtain a first candidate sample according to the sample merging intention data and the corresponding sample merging running data, obtain a second candidate sample according to the sample non-merging intention data and the corresponding sample non-merging running data, and obtain a first training sample according to the first candidate sample and the second candidate sample.

In the embodiment of the disclosure, the first training module 62 is further configured to input a first training sample into the initial driving intention prediction model to obtain a first training result output by the initial driving intention prediction model, obtain a first sample tag of the first training sample to obtain a first training loss of the first training result based on the first sample tag, perform parameter adjustment on the initial driving intention prediction model according to the first training loss, and return to obtain a next first training sample to perform model training on the initial driving intention prediction model after parameter adjustment until training is completed, thereby obtaining a trained candidate driving intention prediction model.

In the embodiment of the disclosure, the second training module 63 is further configured to obtain preset multiple sample driving styles and initial driving style data of a sample vehicle, perform style division on the initial driving style data according to the multiple sample driving styles to obtain respective sample driving style data of the multiple sample driving styles, obtain a candidate driving intention prediction model corresponding to the sample driving style for any one of the sample driving styles, generate a second training sample of the candidate driving intention prediction model according to the sample driving style data under the sample driving style, and train the candidate driving intention prediction model through the second training sample until training is completed to obtain a trained target driving intention prediction model.

In the embodiment of the disclosure, the second training module 63 is further configured to obtain, for any sample driving style, sample driving style intention data and sample driving style driving data under the sample driving style from the sample intention data and the sample driving data, and obtain a second training sample of the candidate driving intention prediction model according to the sample driving style intention data and the sample driving style driving data.

In the embodiment of the disclosure, the second training module 63 is further configured to input a second training sample into the candidate driving intention prediction model to obtain a second training result output by the candidate driving intention prediction model, obtain a second sample tag of the second training sample to obtain a second training loss of the second training result based on the second sample tag, perform parameter adjustment on the candidate driving intention prediction model according to the second training loss, and return to obtain a next second training sample to perform model training on the candidate driving intention prediction model after parameter adjustment until training is completed, thereby obtaining a trained target driving intention prediction model.

In the embodiment of the disclosure, the system further comprises a issuing module, wherein the issuing module is used for acquiring the target driving style of the target vehicle according to any sample driving style, and issuing a target driving intention prediction model corresponding to the sample driving style to the target vehicle in response to the matching of the target driving style and the sample driving style.

The present disclosure provides a training device for a driving intention prediction model, which obtains an initial driving intention prediction model to be trained, and a first training sample constructed by sample intention data and sample driving data, and performs model training on the initial driving intention prediction model through the first training sample to obtain a candidate driving intention prediction model, optionally, obtains sample driving style data of a sample vehicle, and adjusts the candidate driving intention prediction model according to the sample driving style data to obtain a target driving intention prediction model. According to the method and the device, the target driving intention prediction model is obtained through training of the initial driving intention prediction model and adjustment of the candidate driving intention prediction model, so that the target driving intention prediction model can learn the relation between the behavior data of the driver and the driving intention, and therefore the driving intention of the driver is predicted.

The vehicle driving method according to the above embodiments corresponds, and one embodiment of the present disclosure further provides a vehicle driving device, and since the vehicle driving device according to the embodiment of the present disclosure corresponds to the vehicle driving method according to the above embodiments, implementation of the vehicle driving method according to the above embodiments is also applicable to the vehicle driving device according to the embodiment of the present disclosure, and will not be described in detail in the following embodiments.

Fig. 7 is a schematic structural diagram of a vehicle driving device according to an embodiment of the present disclosure, as shown in fig. 7, the vehicle driving device 700 includes a second acquisition module 71, a third acquisition module 72, a prediction module 73, and a control module 74, where:

a second obtaining module 71, configured to obtain a target driving intention prediction model configured on the target vehicle, where the target driving intention prediction model is obtained based on the training device of the driving intention prediction model set forth in the embodiment of fig. 6;

a third acquisition module 72 for acquiring target driving style data of a target vehicle, and target intention data and target travel data of the vehicle in a target time range;

A prediction module 73 for inputting the target intention data, the target travel data, and the target driving style data into a target driving intention prediction model, and outputting a target predicted driving intention of the target vehicle through the target driving intention prediction model;

The control module 74 is configured to perform travel control on the target vehicle based on the target predicted driving intention.

In the embodiment of the disclosure, the device further comprises an updating module, wherein the updating module is used for identifying whether the target driving intention prediction model has corresponding updating data or not, and updating and adjusting the target driving intention prediction model according to the updating data to obtain a new target driving intention prediction model in response to the fact that the corresponding updating data are identified.

In the embodiment of the disclosure, the control module 74 is further configured to perform parallel running control on the target vehicle through the steering lamp control module and the parallel assist module of the vehicle in response to the target predicting the driving intention as the parallel running intention.

In the embodiment of the disclosure, the control module 74 is further configured to, in response to identifying that the parallel running of the target vehicle is finished, finish the parallel running control of the target vehicle, and upload the target predicted driving intention output by the target driving intention prediction model to the cloud.

According to the vehicle running device, the target driving intention prediction model configured on the target vehicle is obtained, the target driving style data of the target vehicle and the target intention data and the target running data of the target vehicle in the target time range are obtained, the target predicted driving intention of the target vehicle is obtained through the target driving intention prediction model according to the target style data, the target intention data and the target running data, and then running control is carried out on the target vehicle according to the target predicted driving intention. In the present disclosure, the driving intention prediction model of the target vehicle obtained by the method of the embodiments of fig. 1 to 2 predicts the driving intention of the target vehicle, which improves the accuracy and efficiency of the driving intention prediction of the target vehicle, and in the case that the auxiliary driving function of the vehicle cannot be started due to the fact that the driving intention of the driver cannot be recognized is avoided in the case that the auxiliary driving function of the vehicle is started by recognizing the driving intention of the vehicle, improves the stability of the driving intention prediction function and the auxiliary driving function of the vehicle, and optimizes the driving experience of the user.

In order to achieve the above embodiments, the present disclosure further provides a vehicle, where the vehicle is used to implement the training method and/or the vehicle driving method of the driving intention prediction model set forth in the above embodiments.

To achieve the above embodiments, the present disclosure also provides an electronic device, a computer-readable storage medium, and a computer program product.

Fig. 8 is a block diagram of an electronic device 800 according to an embodiment of the disclosure, and as shown in fig. 8, the electronic device 800 includes a memory 801, a processor 802, and a computer program stored on the memory 801 and executable on the processor 802, where the processor 802 executes program instructions to implement the training method and/or the vehicle driving method of the driving intention prediction model provided in the foregoing embodiments.

According to the training method of the driving intention prediction model, a first lane changing evaluation parameter set between a first vehicle on a first lane and a second vehicle on a second lane and a second lane changing evaluation parameter set between the second vehicle and a third vehicle on the first lane are obtained, whether the second vehicle meets corresponding target lane changing conditions or not is identified according to the first lane changing evaluation parameter set and/or the second lane changing evaluation parameter set, and when the second vehicle meets the target lane changing conditions, the second vehicle is controlled to move from the second lane changing to the first lane. In the method, whether the second vehicle meets the target lane change condition or not is evaluated through the first lane change evaluation parameter set and/or the second lane change evaluation parameter set, evaluation is performed based on a plurality of parameters between the first vehicle and the second vehicle and/or a plurality of parameters between the second vehicle and the third vehicle, the number of parameters used in the evaluation process is increased, the evaluation parameter range is enlarged, the accuracy of an evaluation result of whether the second vehicle meets the target lane change condition or not is improved, the accuracy of a judgment result of whether the second vehicle can perform lane change or not is further improved, the safety of the vehicle during lane change driving is optimized, and driving experience of a user is further optimized.

Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuit systems, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems On Chip (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include being implemented in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be a special or general purpose programmable processor, operable to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

Program code for carrying out the methods themselves may be written in any combination of one or more programming languages. These program code may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus such that the program code, when executed by the processor or controller, causes the functions/operations specified in the flowchart and/or block diagram to be implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.

In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user, for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback), and input from the user may be received in any form, including acoustic input, speech input, or tactile input.

The systems and techniques described here can be implemented in a computing system that includes a background component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a grid browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such background, middleware, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication grid). Examples of communication grids include Local Area Networks (LANs), wide Area Networks (WANs), the internet, and blockchain grids.

The computer system may include a client and a server. The client and server are typically remote from each other and typically interact through a communications grid. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also called a cloud computing server or a cloud host, and is a host product in a cloud computing service system, so that the defects of high management difficulty and weak service expansibility in the traditional physical hosts and VPS service ("Virtual PRIVATE SERVER" or simply "VPS") are overcome. The server may also be a server of a distributed system or a server that incorporates blockchains.

In the description of the present specification, a description referring to terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present disclosure. In this specification, schematic representations of the above terms are not necessarily directed to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, the different embodiments or examples described in this specification and the features of the different embodiments or examples may be combined and combined by those skilled in the art without contradiction.

Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include at least one such feature. In the description of the present disclosure, the meaning of "a plurality" is at least two, such as two, three, etc., unless explicitly specified otherwise.

Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing specific logical functions or steps of the process, and additional implementations are included within the scope of the preferred embodiment of the present disclosure in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the embodiments of the present disclosure.

Logic and/or steps represented in the flowcharts or otherwise described herein, e.g., a ordered listing of executable instructions for implementing logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium would include an electrical connection (an electronic device) having one or more wires, a portable computer diskette (a magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium may even be paper or other suitable medium upon which the program is printed, as the program may be electronically captured, via, for instance, optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.

It should be understood that portions of the present disclosure may be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, the various steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented in hardware as in another embodiment, may be implemented using any one or combination of techniques known in the art, discrete logic circuits with logic gates for implementing logic functions on data signals, application specific integrated circuits with appropriate combinational logic gates, programmable Gate Arrays (PGAs), field Programmable Gate Arrays (FPGAs), etc.

Those of ordinary skill in the art will appreciate that all or part of the steps carried out in the method of the above-described embodiments may be implemented by a program to instruct related hardware, and the program may be stored in a computer readable storage medium, where the program when executed includes one or a combination of the steps of the method embodiments.

Furthermore, each functional unit in the embodiments of the present disclosure may be integrated in one processing module, or each unit may exist alone physically, or two or more units may be integrated in one module. The integrated modules may be implemented in hardware or in software functional modules. The integrated modules may also be stored in a computer readable storage medium if implemented as software functional modules and sold or used as a stand-alone product.

The above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, or the like. Although embodiments of the present disclosure have been shown and described above, it will be understood that the above embodiments are illustrative and not to be construed as limiting the present disclosure, and that variations, modifications, alternatives, and variations may be made to the above embodiments by one of ordinary skill in the art within the scope of the present disclosure.

It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps recited in the present disclosure may be performed in parallel, sequentially, or in a different order, provided that the desired results of the disclosed aspects are achieved, and are not limited herein.

The above detailed description should not be taken as limiting the scope of the present disclosure. It will be apparent to those skilled in the art that various modifications, combinations, sub-combinations and alternatives are possible, depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present disclosure are intended to be included within the scope of the present disclosure.

Claims

1. A method for training a driving intention prediction model, characterized in that the method comprises:

Acquire an initial driving intention prediction model to be trained, and acquire sample intention data and sample driving data of a sample vehicle to obtain a first training sample of the initial driving intention prediction model;

Training the initial driving intention prediction model according to the first training sample until the training is completed to obtain a trained candidate driving intention prediction model;

Acquiring sample driving style data of the sample vehicle, and adjusting the candidate driving intention prediction model according to the sample driving style data to obtain an adjusted target driving intention prediction model;

The acquiring of sample intention data and sample driving data of a sample vehicle to obtain a first training sample of the initial driving intention prediction model includes:

Performing line merging intention recognition on the sample intention data to obtain sample line merging intention data and sample non-line merging intention data from the sample intention data;

Acquire, from the sample driving data, sample merging driving data corresponding to the sample merging intention data, and sample non-merging driving data corresponding to the sample non-merging intention data;

According to the sample merging intention data and the corresponding sample merging driving data, construct a training positive sample of the initial driving intention prediction model, and mark it as a first candidate sample; and according to the sample non-merging intention data and the corresponding sample non-merging driving data, construct a training negative sample of the initial driving intention prediction model, and mark it as a second candidate sample;

The first training sample is obtained according to the first candidate sample and the second candidate sample.

2. The method according to claim 1, characterized in that the initial driving intention prediction model is trained according to the first training sample until the training is completed to obtain a trained candidate driving intention prediction model, comprising:

Inputting the first training sample into the initial driving intention prediction model to obtain a first training result output by the initial driving intention prediction model;

Obtaining a first sample label of the first training sample to obtain a first training loss of the first training result based on the first sample label;

The parameters of the initial driving intention prediction model are adjusted according to the first training loss, and the next first training sample is returned to continue model training on the initial driving intention prediction model after the parameters are adjusted until the training is completed, so as to obtain the trained candidate driving intention prediction model.

3. The method according to claim 1, characterized in that the acquiring the sample driving style data of the sample vehicle and adjusting the candidate driving intention prediction model according to the sample driving style data to obtain an adjusted target driving intention prediction model comprises:

Acquire a plurality of preset sample driving styles and initial driving style data of the sample vehicle, and perform style classification on the initial driving style data according to the plurality of sample driving styles to obtain sample driving style data of each of the plurality of sample driving styles;

For any sample driving style, obtaining a candidate driving intention prediction model corresponding to the sample driving style, and generating a second training sample of the candidate driving intention prediction model according to the sample driving style data under the sample driving style;

The candidate driving intention prediction model is trained using the second training sample until the training is completed, thereby obtaining the trained target driving intention prediction model.

4. The method according to claim 3, characterized in that, for any sample driving style, obtaining a candidate driving intention prediction model corresponding to the sample driving style, and generating a second training sample of the candidate driving intention prediction model according to the sample driving style data under the sample driving style, comprises:

For any sample driving style, acquiring sample driving style intention data and sample driving style driving data under the sample driving style from the sample intention data and the sample driving data;

The second training sample of the candidate driving intention prediction model is obtained according to the sample driving style intention data and the sample driving style travel data.

5. The method according to claim 3, characterized in that the step of training the candidate driving intention prediction model by using the second training sample until the training is completed to obtain the trained target driving intention prediction model comprises:

Inputting the second training sample into the candidate driving intention prediction model to obtain a second training result output by the candidate driving intention prediction model;

Obtaining a second sample label of the second training sample to obtain a second training loss of the second training result based on the second sample label;

The parameters of the candidate driving intention prediction model are adjusted according to the second training loss, and the next second training sample is returned to continue model training on the candidate driving intention prediction model after the parameter adjustment until the training is completed, so as to obtain the trained target driving intention prediction model.

6. The method according to any one of claims 3 to 5, characterized in that the step of training the candidate driving intention prediction model according to the second training sample until the training is completed and a trained target driving intention prediction model is obtained comprises:

For any sample driving style, obtain a target driving style of a target vehicle;

In response to the target driving style matching the sample driving style, a target driving intention prediction model corresponding to the sample driving style is sent to the target vehicle.

7. A vehicle driving method, characterized in that the method comprises:

Acquire a target driving intention prediction model configured on the target vehicle, wherein the target driving intention prediction model is obtained based on the training method of the driving intention prediction model according to any one of claims 1 to 6;

Acquiring target driving style data of the target vehicle, and target intention data and target travel data of the vehicle within a target time range;

Inputting the target intention data, the target driving data and the target driving style data into the target driving intention prediction model, and outputting the target predicted driving intention of the target vehicle through the target driving intention prediction model;

Based on the target predicted driving intention, driving control is performed on the target vehicle.

8. The method according to claim 7, characterized in that the method further comprises:

Identify whether corresponding update data exists for the target driving intention prediction model;

In response to identifying the existence of the update data, the target driving intention prediction model is updated and adjusted according to the update data to obtain a new target driving intention prediction model.

9. The method according to claim 7, characterized in that the driving control of the vehicle based on the target predicted driving intention comprises:

In response to the target predicted driving intention being a lane-merging driving intention, the target vehicle is controlled to be lane-merging through the turn signal control module and the lane-merging auxiliary module of the vehicle.

10. The method according to claim 9, characterized in that after the vehicle is controlled based on the target driving intention, the method further comprises:

In response to identifying that the target vehicle has ended its lane-merging driving, the lane-merging driving control of the target vehicle is ended, and the target predicted driving intention output by the target driving intention prediction model is uploaded to the cloud.

11. A training device for a driving intention prediction model, characterized in that the device comprises:

A first acquisition module is used to acquire an initial driving intention prediction model to be trained, and acquire sample intention data and sample driving data of a sample vehicle to obtain a first training sample of the initial driving intention prediction model;

A first training module, used for training the initial driving intention prediction model according to the first training sample until the training is completed to obtain a trained candidate driving intention prediction model;

a second training module, configured to obtain sample driving style data of the sample vehicle, and adjust the candidate driving intention prediction model according to the sample driving style data to obtain an adjusted target driving intention prediction model;

12. A vehicle driving device, characterized in that the device comprises:

A second acquisition module is used to acquire a target driving intention prediction model configured on the target vehicle, wherein the target driving intention prediction model is obtained based on the training device of the driving intention prediction model according to claim 11;

A third acquisition module is used to acquire target driving style data of the target vehicle, and target intention data and target travel data of the vehicle within a target time range;

A prediction module, used for inputting the target intention data, the target driving data and the target driving style data into the target driving intention prediction model, and outputting the target predicted driving intention of the target vehicle through the target driving intention prediction model;

A control module is used to control the driving of the target vehicle based on the target predicted driving intention.

13. A vehicle, characterized in that the vehicle is used to implement the method described in any one of claims 1 to 6 and/or claims 7 to 10.

14. An electronic device, comprising:

processor;

a memory for storing executable instructions for the processor;

The processor is configured to execute instructions to implement the method according to any one of claims 1-6 and/or claims 7-10.

15. A computer-readable storage medium, when instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the method according to any one of claims 1 to 6 and/or claims 7 to 10.