CN103753585A - Method for intelligently adjusting manipulator and grasping force on basis of visual image analysis - Google Patents

Abstract

The invention relates to a manipulator and a method for adjusting its grasping force. The manipulator also includes a visual image acquisition system and a database; the visual image acquisition system captures a 3D solid image of an object to be grasped by the manipulator, and simultaneously scans the acquired The size information of the object is obtained and sent to the central processing unit; the database stores the image data of various items and the parameters of density, roughness and friction coefficient of the material; the central processing unit combines the collected 3D solid images with the database Compare and judge the image data in, determine the type, size and volume of the grasped object, and call out the density, roughness and friction coefficient parameters of the material of the object. The central processing unit outputs the grasping force information to the control unit according to the above parameters. The control unit controls the grasping force and lifting force of the mechanical arm. The present invention obtains a three-dimensional solid image, which can more accurately calculate the grasping force required to be output by the mechanical arm.

Description

A method for intelligent adjustment of manipulator and gripping force based on visual image analysis

the

technical field

The invention relates to a manipulator and an adjustment method for its gripping force, in particular to an intelligent adjustment method for a manipulator and its gripping force based on visual image analysis. the

the

Background technique

With the development of science and technology, the level of robot technology has been continuously improved, and its application fields have also been continuously expanded, and it has also solved many problems in people's lives. The domestic manufacturing industry is facing the problem of employing labor: labor-intensive manufacturing enterprises need to hire a large number of laborers, but after the annual long holiday, some front-line personnel always lose some employees, which sometimes interferes with normal production. Frequent operator replacement will affect product quality and increase The rate of equipment damage and potential safety hazard accidents increased. Nowadays, many workers are unwilling to do heavy-duty work on the production line. In addition, labor costs continue to increase. These issues are driving companies to continuously introduce robots. In the future, there will be many applications for home robots, such as cleaning and caring for the elderly. According to the sixth national census in 2010, the number of households with elderly population aged 60 and above was 123 million, of which 43 million were empty-nest families (children went out to work or study, and the elderly lived alone). With the continuous maturity of service robot technology, there will be more and more applications of intelligent robots for household cleaning and medical care. But the key problem at present is: people can judge based on experience how much grasping force and lifting force are needed to grasp an object, so as to ensure that the object to be grasped is grasped and lifted, which involves two problems: first, it will not slip and fall, How much is the second lifting force. Moreover, it can automatically increase the grasping force and lifting force through the brain response according to changes in the situation during the snatch process, such as slippery, heavy weight, etc. This is the advantage of human intelligence. At present, manipulators are still unable to do this, but use various sensor technologies to develop in this direction. For example: when the manipulator grabs some items (body), such as smooth cups, biscuits, cigarettes, mineral water and other soft-packed items that are afraid of being crushed, how to ensure that the manipulator is grabbing them, such as a cylinder filled with water It is a difficult problem to get the grip just right for a shaped, smooth glass without overwhelming the grip and crushing the quilt. At the same time, it is also necessary to be able to judge its weight according to the volume and different materials, and to adjust the grasping and moving (lifting, etc.) force is another key problem. This patent solves these two major problems based on the visual image function. the

In addition, in terms of home robots, it is of great significance and role, especially in dealing with some common but ever-changing items. For example, holding a cup for a while, a paper towel for a while, a child’s rubber toy for a while, and a fruit and banana for a while, the home robot needs to distinguish these commonly used items, so as to apply force reasonably. the

the

Contents of the invention

The purpose of the present invention is to provide an intelligent adjustment method of the manipulator and its grasping force based on visual image analysis, which has the characteristics of low cost and sensitive response, and can be used for the grasping activities of household robots and industrial production. the

Technical scheme of the present invention is as follows:

A kind of manipulator based on visual image analysis, described manipulator comprises central processing unit, control unit and mechanical arm, is characterized in that:

The manipulator also includes a visual image acquisition system and a database; the visual image acquisition system captures the 3D solid image of the object to be grasped by the manipulator, and scans the size information of the collected object at the same time to obtain the volume and send it to the central processing unit ;

The database stores image data of various items and parameters of material density, roughness, and coefficient of friction;

The central processing compares and judges the collected 3D solid image with the image data in the database, determines the type, size and volume of the grasped object, and calls out the density, roughness, and friction coefficient parameters of the material of the object, and the central processing The controller outputs information on the grasping force to the control unit according to the above parameters, and the control unit controls the grasping force and lifting force of the mechanical arm.

Further, the visual image acquisition system includes an optical scanning system, and the optical scanning system includes two CCD cameras and a grating generator;

The two CCD cameras are installed on the head of the robot at different angles, and the grating generator projects multiple groups of grating stripes onto the object surface, and the two CCD cameras at different angles simultaneously photograph the stripe patterns on the object surface, and input the stripe images into the central processing unit.

Further, a displacement sensor is installed on the end of the mechanical finger, and when a slight slippage is detected, the central processing unit increases the gripping force according to the degree of slippage. the

Further, the database is continuously updated with new item image data and material information, and remembers the features of the captured items to be called when the next repeated capture is performed. the

the

A method for intelligently adjusting the grasping force of a manipulator, the steps of the method are as follows:

(1) The visual image acquisition system captures the 3D solid image of the object to be grasped by the manipulator, and scans the size information of the acquired object at the same time, obtains the volume size, and transmits it to the central processing unit;

(2) The central processor compares and judges the collected entity image with the database, and determines the material density, roughness, friction coefficient and other parameters of the grasped object, and the central processor outputs the grasping force information to the control unit according to the above parameters. The control unit controls the grasping force and lifting force of the mechanical arm.

Further, in the step (1), the optical scanning step method of the visual image acquisition system is as follows:

The grating generator in the optical three-dimensional scanning system projects multiple groups of grating fringes onto the object surface, and two CCD cameras at different angles simultaneously capture the fringe patterns on the object surface, and input the fringe images into the computer, and use the phase method according to the fringe curvature changes The three-dimensional point cloud data of the space coordinates (X, Y, Z) of each point on the surface of the object can be accurately calculated by using trigonometry and trigonometry, and the three-dimensional space distance and size of the object can be judged; the data of multiple shots can be accurately captured by using computer control, multi-angle shooting Stitched together, the shooting result is a 3D stereoscopic image of the object, the size and volume information of the object is obtained, and the object can be identified according to the shape characteristics of the object; when it is identified as a certain object, the density of the object in the database is called, rough According to parameters such as degree and coefficient of friction, the central processing unit outputs information on the grasping force to the control unit, and the control unit controls the grasping force and lifting force of the mechanical arm.

Compared with the prior art, the technical solution of the present invention has the following beneficial effects:

(1) The technical solution of the present invention obtains a three-dimensional solid image, so compared with the two-dimensional image obtained by the ordinary image acquisition system, it has a more accurate object discrimination rate, so that it can more accurately calculate the grasp output required by the manipulator. Grip.

(2) Ordinary image acquisition systems cannot obtain the external dimensions and volume information of objects, but the scanning function of this visual image acquisition system can accurately obtain these information. the

the

Description of drawings

Fig. 1 is a functional principle block diagram of the patent of the present invention. the

Figure 2 is a schematic diagram of the structure of the robot. the

Fig. 3 is a schematic diagram of the structure of the head of the robot. the

Among them, 1-robot head; 2-manipulator, 3-optical grating generator, 4-CCD camera, among which, 3 and 4 form a visual image acquisition system. the

the

Detailed ways

Taking a water glass as an example, first determine that it is a glass: by comparing the captured image with the existing images in the database, determine the glass water glass; then determine the position of the water, and then determine the size, size and volume of the glass, and calculate The approximate volume of the water is produced, and then the weight of the cup is calculated. According to the information stored in the database, the central processing unit calculates the amount of gripping force required to ensure that the friction force can prevent the cup from slipping when grabbing it. Then through the controller, control the movement of the manipulator and clamp (catch) the smooth glass with a suitable force. the

For some commonly used standard items, such as cigarettes, mineral water, etc., after the judgment is made, the clamping force values established for these items in the database can be directly called. the

Its exterior is scanned by an optical scanner, the volume is calculated by the CPU, first gravity is calculated, and then friction is calculated to determine the grasping force that should be applied when grasping an object. the

At the same time, there is a displacement sensor on the end of the mechanical finger, but when a small slip is detected, the gripping force is increased according to the degree of slip. the

At the same time, when grabbing other items, the material can be determined through image recognition and comparison with the database, and the volume can be calculated, and the weight can be obtained by multiplying the density of the material by the volume. The weight of the body and the required clamping force can be calculated by calling out the friction coefficient of the material, which can realize artificial intelligence adjustment. the

The database can expand the item information, and the system has a self-learning function, which can memorize the characteristics of the captured items and call them for the next repeated capture, thereby increasing the speed. However, when the shape of the object is complex, the vision system can scan 360° around the object, so that the complete three-dimensional entity of the part can be obtained, and its volume can be accurately calculated. When the manipulator 2 is ready to grab the object, it first takes pictures of the object through the image acquisition system, judges the material by comparing with the database, and obtains the relevant parameters of the corresponding material, such as density, roughness, etc. the

Before the robot grabs the item, it first scans the item. The robot head 1 is equipped with two CCD cameras 4 with different angles. The grating generator 3 in the optical three-dimensional scanning system projects multiple groups of grating fringes onto the object surface, and two CCD cameras at different angles simultaneously photograph the fringe patterns on the object surface, and input the fringe images into the computer, and use phase The three-dimensional point cloud data of the spatial coordinates (X, Y, Z) of each point on the surface of the object can be accurately calculated by using the method and trigonometry, and the three-dimensional space distance and size of the object can be judged. Using computer control, multi-angle shooting and accurate stitching together of multiple shooting data, the shooting result is a 3D stereoscopic image of the object. In this way, information such as object size and volume can be obtained. And according to the appearance characteristics of the item, it can be compared with the graphic data in the database to identify which item it is. Once it is identified as a certain item, the density of the item in the database can be called to calculate the weight. For example: when it is judged as a bottle of mineral water, the volume of the plastic bottle is known and the height of the water can be observed, the total weight can be calculated according to the volume and density of the plastic and water, and finally calculated according to the friction coefficient of the plastic bottle Appropriate clamping force and lifting force, etc. the

Through the calculation of the processor, the weight of the object and the friction force of grasping are obtained, so as to determine the grasping force and lifting force of the manipulator, and avoid damage to the object due to excessive grasping force. the

 The calculation results are converted into control commands by the control unit and sent to the manipulator to guide the manipulator to complete the action. the

The scanning method can be used for scanning soft and fragile objects and occasions where contact scanning is difficult or not allowed, and the scanning accuracy is high. The 3D stereoscopic image of the object obtained through the image acquisition system is used to distinguish the material, which is more accurate and precise than the two-dimensional image acquisition. In this way, the deformation of the object can also be monitored in real time, and real-time instructions can be transmitted to the manipulator, so that the manipulator can adjust the force according to the shape of the object, which has the characteristics of high sensitivity, safety and reliability. the

Claims (6)

1. a kind of manipulator based on visual image analysis, described manipulator comprises central processing unit, control unit and mechanical arm, is characterized in that: The manipulator also includes a visual image acquisition system and a database; the visual image acquisition system captures the 3D solid image of the object to be grasped by the manipulator, and scans the size information of the collected object at the same time to obtain the volume and send it to the central processing unit ; The database stores image data of various items and parameters of material density, roughness, and coefficient of friction; The central processing compares and judges the collected 3D solid image with the image data in the database, determines the type, size and volume of the grasped object, and calls out the density, roughness, and friction coefficient parameters of the material of the object, and the central processing The controller outputs information on the grasping force to the control unit according to the above parameters, and the control unit controls the grasping force and lifting force of the mechanical arm. 2. The manipulator according to claim 1, wherein the visual image acquisition system includes an optical scanning system, and the optical scanning system includes two CCD cameras and a grating generator; The two CCD cameras are installed on the head of the robot at different angles, and the grating generator projects multiple groups of grating stripes onto the object surface, and the two CCD cameras at different angles simultaneously photograph the stripe patterns on the object surface, and input the stripe images into the central processing unit. 3. The manipulator according to claim 1, characterized in that: A displacement sensor is installed on the end of the mechanical finger, and when a slight slip is detected, the central processing unit increases the grasping force according to the degree of slip. 4. The manipulator according to claim 1, characterized in that: The database is constantly updated with new item image data and material information, and remembers the features of the captured items for calling when the next repeated capture. 5. The method for intelligently adjusting the grasping force of the manipulator according to claim 1, the method steps are as follows: (1) The visual image acquisition system captures the 3D solid image of the object to be grasped by the manipulator, and scans the size information of the acquired object at the same time, obtains the volume size, and transmits it to the central processing unit; (2) The central processor compares and judges the collected entity image with the database, and determines the material density, roughness, friction coefficient and other parameters of the grasped object, and the central processor outputs the grasping force information to the control unit according to the above parameters. The control unit controls the grasping force and lifting force of the mechanical arm. 6. The method according to claim 4, characterized in that: In the step (1), the optical scanning step method of the visual image acquisition system is as follows: The grating generator in the optical three-dimensional scanning system projects multiple groups of grating fringes onto the object surface, and two CCD cameras at different angles simultaneously capture the fringe patterns on the object surface, and input the fringe images into the computer, and use the phase method according to the fringe curvature changes The three-dimensional point cloud data of the space coordinates (X, Y, Z) of each point on the surface of the object can be accurately calculated by using trigonometry and trigonometry, and the three-dimensional space distance and size of the object can be judged; the data of multiple shots can be accurately captured by using computer control, multi-angle shooting Stitched together, the shooting result is a 3D stereoscopic image of the object, the size and volume information of the object is obtained, and the object can be identified according to the shape characteristics of the object; when it is identified as a certain object, the density of the object in the database is called, rough According to parameters such as degree and coefficient of friction, the central processing unit outputs information on the grasping force to the control unit, and the control unit controls the grasping force and lifting force of the mechanical arm.

Images