[go: up one dir, main page]

WO2018109725A1 - Procédé et dispositif de détection de chaleurs chez un ruminant - Google Patents

Procédé et dispositif de détection de chaleurs chez un ruminant Download PDF

Info

Publication number
WO2018109725A1
WO2018109725A1 PCT/IB2017/057967 IB2017057967W WO2018109725A1 WO 2018109725 A1 WO2018109725 A1 WO 2018109725A1 IB 2017057967 W IB2017057967 W IB 2017057967W WO 2018109725 A1 WO2018109725 A1 WO 2018109725A1
Authority
WO
WIPO (PCT)
Prior art keywords
animal
information
estrus
behavior
feeding
Prior art date
Application number
PCT/IB2017/057967
Other languages
English (en)
Inventor
Leonardo Luis GIOVANINI
Sebastián Rodrigo VANRELL
Diego Humberto MILONE
Julio Ricardo GALLI
Hugo Leonardo RUFINER
José Omar CHELOTTI
Original Assignee
Consejo Nacional De Investigaciones Científicas Y Técnicas (Conicet)
Universidad Nacional Del Litoral
Universidad Nacional De Rosario
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Consejo Nacional De Investigaciones Científicas Y Técnicas (Conicet), Universidad Nacional Del Litoral, Universidad Nacional De Rosario filed Critical Consejo Nacional De Investigaciones Científicas Y Técnicas (Conicet)
Publication of WO2018109725A1 publication Critical patent/WO2018109725A1/fr

Links

Classifications

    • A—HUMAN NECESSITIES
    • A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
    • A01K—ANIMAL HUSBANDRY; AVICULTURE; APICULTURE; PISCICULTURE; FISHING; REARING OR BREEDING ANIMALS, NOT OTHERWISE PROVIDED FOR; NEW BREEDS OF ANIMALS
    • A01K29/00—Other apparatus for animal husbandry
    • A01K29/005—Monitoring or measuring activity
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61D—VETERINARY INSTRUMENTS, IMPLEMENTS, TOOLS, OR METHODS
    • A61D17/00—Devices for indicating trouble during labour of animals ; Methods or instruments for detecting pregnancy-related states of animals
    • A61D17/002—Devices for indicating trouble during labour of animals ; Methods or instruments for detecting pregnancy-related states of animals for detecting period of heat of animals, i.e. for detecting oestrus

Definitions

  • This invention provides a process and a non-invasive device for the detection of estrus in bovine cattle by means of the detection, identification, and analysis for a period of time of any changes in feeding, in the group and physical behavior, based on the recording and analysis of movements, position and sounds made by the animal when feeding and their relation with the estrus periods .
  • cows for both, production of calves or milk
  • it is relevant to improve the use of the resources available along the entire cycle to increase the reproductive efficiency of the cattle.
  • Each cow that fails to become pregnant turns into an unproductive resource that needs to be maintained, thus decreasing its productive efficiency.
  • an efficient reproductive management is decisive for the productivity of husbandry systems. For that purpose, it is important to determine when a cow is in estrus, as this allows determining the right time for the natural mounting activity by bulls (servicing on farm) or artificial insemination by farmers.
  • the most important factor is to know its symptoms.
  • the most reliable symptom for a cow in estrus is that is allows being mounted (it remains quiet) by bulls or other cows. Such characteristic symptom may last from four to six seconds. Approximately, a cow in estrus is mounted three or four times every two hours, and this behavior (estrus or heat) is visible during six to twelve hours, depending on various environmental and animal' s factors (the time of the year) .
  • the methods used to determine when a cow is in estrus include
  • teaser males i.e. Animals that may not copulate and are sexually sensitive but cannot impregnate the female because they cannot release semen (operated males or androgenized females) and, therefore, when a female is mounted, it is not serviced, but it may be noticed by the farm worker;
  • Devices for the detection of activity using mechanical or electronic movement sensors located at the cow limbs (or pedometers) or on its neck (collars) to measure the increase in their activity during estrus.
  • Pedometers measure the number of steps taken by the animal while the necklaces measure the animal's general movement. When the cow changes its level of physical activity, the device provides some indication of the event that may be noticed.
  • estrus detection devices and methods have been developed to detect the mounting of animals or their physical activity change
  • measuring the physical activity level by the analysis of the animal' s body movement is a known and non-invasive technique that may be implemented with commercial sensors (HR Heatime Systems and SCR Pro Heatime, DeLaval ALPRO system) .
  • HR Heatime Systems and SCR Pro Heatime, DeLaval ALPRO system The trend to gather together with attempts to mount other animals in such groups has not been taken into account for the automation.
  • one of the most evident changes in the state of estrus may be noticed in the changes in the animal's feeding behavior. In this sense, the amount of matter ingested and the time spent by the animal ruminating decrease.
  • the state of estrus may be detected based on the analysis of changes in the feeding behavior, which may be accomplished through the analysis of the sound produced by the animal during forage rumination, both during ingestion and during rumination.
  • This is a known and non-invasive technique that may be used with commercial sensors and known methods, and may be adapted for their use in automatic estrus detection.
  • the only implementations available for this approach consist in measuring changes in rumination periods, which is combined with the physical activity level. When a cow increases the activity and decreases rumination (associated to a lower feed level) with respect to previous days, that animal would be in estrus.
  • the devices for estrus detection are mainly electronic devices using movement sensors, and are mounted on the animal limbs (pedometers) or on the animal's neck (collars) to measure only changes in their activity level, but ignoring changes in feeding and group behaviors of such animals.
  • Pedometers measure the number of steps taken, but are not easy to install and may provide very inaccurate information.
  • daily activity changes may be due to other reasons (changes of plots or parcels of land) and not to the estrus period; therefore, "false positives” are indicated.
  • environmental conditions frequently cause a failure in such devices and/or may cause damages and pain to the animal.
  • the devices mounted on the animal's neck measure the general movement of the animal, but they are also sensitive to other activities (eating, smelling and playing) , so they frequently provide inaccurate information .
  • a wide range of sensors have been used to detect such activity.
  • a technique known to measure the activity uses a glass vial containing electrodes and a drop of mercury. When there is movement, the mercury covers the electrodes and the impedance between the electrodes decreases.
  • Another methodology used to measure the activity consists in a metal ball that creates and electromagnetic field when the location is changed.
  • inertial sensors based on MEMS technology have been used to measure the activity of animals.
  • gravity affects such sensors, so that the sensor inclination changes and the reading changes with the movement, thus requiring extra-processing for their re- calibration .
  • Patent US 4,846,106 describes an electronic device mounted on the rear part of the cow that turns on when the animal is mounted by another animal .
  • Patent US 4,247,758 describes a device that measures and indicates the number of movements the animal has made for detecting estrus.
  • Patent US 4,618,861 describes a device that measures an animal's activity, using the device movement as a source of energy .
  • Patent US 5,111,799 describes an electronic switch mounted on the rear part of the cow that turns on when the animal is mounted by another animal .
  • Patent US 5,542,431 describes a system to determine the number of times an animal is mounted using a switch located on the rear part of the animal.
  • Patent US 5,881,673 describes a system to determine the number of times and the time an animal is mounted using a switch located at the area of the tail of the animal.
  • Patent US 7.083.75 Bl describes a device using a switch located at the area of the tail of the animal to detect the number of times an animal is mounted.
  • Patent US 7,137,359 Bl describes a device using a pressure sensor to detect when an animal mounts another animal.
  • Patent US 7,230,535 B2 describes an electronic switch mounted on the cow that turns on when the animal mounts another animal. This information is used to determine when the animal is in estrus and it is then submitted to the user.
  • Patent US 7,878,149 B2 describes a device and a method using an inclinometer to measure the animal' s head movement and its activity, and then, it is detected when the animal is in estrus by means of statistical analysis.
  • Patent US 7,927,287 B2 describes a method and a device to determine the reproductive status of the animals based on the identification of feeding events (estrus, mounting, pregnancy, delivery, among others) .
  • the device provides an indicative signal when an animal is in estrus using an adhesive layer that is attached to the animal's protuberance.
  • On the upper surface of the adhesive layer there is an indicative layer that is covered by a coating layer.
  • the coating layer is adapted for its elimination after mounting, exposing the indicative layer and generating an indication of mounting.
  • Patent US 7,992,521 B2 describes a layer for automatic detection of mating between animals.
  • Such device includes a detector for mating attempts and an electronic identifier that is inserted in the female' s body and is actuated by the detector.
  • Such identifier includes a transceiver with an electronic tag. The transceiver may read the animal' s electronic identifier may read the electronic identifier of the animal attempting to mount another animal. Such device may store representative information about the mounting attempt and then send it by the transceiver.
  • Patent US 8,066,179 describes a method and a system that allows detecting and identifying when an animal is mounted by another based on the position of such animals, using radio- frequency identification devices (RFID) .
  • RFID radio- frequency identification devices
  • Patent US 8,420,398 B2 describes a method for the determination of the estrus cycle phase, where the animal is at the time of obtaining the biological sample.
  • the biological sample is combined with at least a flavonoid pigment that reacts with the sample on a hydrophobic surface to provide a color response.
  • the estrus cycle phase has a color response that may be distinguished without requiring any assistance.
  • Patent 8,662,021 B2 describes a device and a method using an electronic and mechanical sensor located on one of the limbs to determine the times when the animal is standing or lying.
  • Patent US 8,949,034 B2 describes a method for determining when an animal is in estrus and its health condition based on an analysis of its location and movements inside barns.
  • the method suggested estimates and analyzes the movement characteristic parameters of the animals, such as the percentage of time that the animals are walking and the number of times the animals moved at a predefined distance over a given period, among others.
  • the method suggested supposes that a system is available for measurement and recording of the displacement of animals inside the yard using RFID devices and ultra broad-band communication systems .
  • Patent US 8,979,757 B2 describes a method and a system for the control of the cattle status, comprising a number of sensors for detecting the relevant parameters of the behavior of animals. The information recorded is sent by a wireless device mounted on the animal to a central processor, which determines the animal condition when it is about to deliver, the fertility status and other health conditions of the animal.
  • Patent publications US20140107434 and US20150202033 Al describe a method for the detection of estrus using information about the location of the animals inside a barn. Such information includes the location of the animals at regular intervals over a period of time. This information is used for creating behavior indicators (time standing still, time lying, drinking time, counting of mounting events, among others), which is used for the determination of the animal's reproductive condition.
  • Patent publication US 20140121558 Al describes a device or set of devices used for remotely, continuous and autonomous measuring biological parameters and for locating animals. This allows the early detection of illnesses and other biological conditions, in particular, epidemic outbreaks or robbery of cattle, thus facilitating the early reaction of owners and/or authorities.
  • Patent publication US 20160165851 Al describes a device attached to the animal' s neck using an accelerometer to create indicative signals of the animal' s body and head movement during fifteen-minute periods. Such signals are analyzed to detect the status of animals: resting, feeding, and three physical activity levels. The data indicating the animal's statuses are saved in the device and periodically transmitted to a server.
  • This invention provides a process and a non-invasive device for the detection of estrus in animals by gathering information related to the physical activity, feeding, and group behavior of the animal. Such information is combined to reduce the impact of habits and anomalous situations in the detection process.
  • the process and device of this invention has the advantage of using and linking all the information available of the different aspects of the animal's behavior in a non-invasive fashion. Such capability improves the information interpretation obtained by the sensors, thus achieving a reduction in the estimation of errors.
  • the information integration is essential, as each of the information sources used separately has limitations and issues:
  • Ingestive behavior changes in the feeding behavior can also be caused by illnesses or diet changes.
  • the process for the detection of estrus in a ruminant animal which is the subject matter of this invention comprises the following steps:
  • Such invention process also comprises a database of typical behavior, and such database is continuously updated. It also comprises the stage of delivery to the remote receiver of a list of the movement detected, the information related to the periods of feeding, the information about the group behavior and the identification of the animal's status. On the other hand, it comprises a stage to deliver a notice when the animal is in estrus, and at least one of such steps is performed by a device that is placed on the animal's head, such that the animal' s head inclination may be measured.
  • step "b” of detecting, sorting and quantifying the data indicating the feeding activity comprises a stage for measuring the animal's head inclination with respect to the ground and its movements; and such step “b", of detecting, sorting and quantifying the data indicating the feeding activity comprises the recording, processing and analysis of sounds and vibrations produced by the animal when feeding.
  • step “c” comprises a stage for measuring and processing translational accelerations, the rotation speeds and the measurement of angles with respect to the magnetic north of the Earth in up to three orthogonal directions.
  • step “d” comprises a stage for measuring and processing of vertical translational accelerations and the identification of the mounting animal through a near-field communication system.
  • step "c” comprising a stage of identification of the females and bulls in estrus that are at a given distance from the female considered by a near-field communication system and GPS devices .
  • Another purpose of this invention is a device for the detection of estrus in a ruminant female comprising: a. At least one inertial measurement unit (IMU) and GPS for measuring the movement and position of such animal during a period of time, in which the activity level is indicated by parameters that indicate the movement signal resulting from the composition of signals produced by the IMU;
  • IMU inertial measurement unit
  • At least one sound sensor for the detection during a period of time, of data indicating such animal's feeding activity
  • NFC near-field communications
  • such device in a preferred mode, also, comprise at least an embedded system (based on microprocessors, FPGA or ASIC) to process the signals received, compose the IMU signals to correct errors and improve the accuracy of the data resulting from the position and movement, extract the typical activity level of such animal based on such movement data and identify the abnormal behavior, indicating the estrus based on an analysis of the parameters indicating the movement data.
  • an embedded system based on microprocessors, FPGA or ASIC
  • It also comprises at least one embedded system (based on microprocessors, FPGA or ASIC) to process the sound signals or vibrations received, filtering sounds and/or disturbances, and extracting relevant information about the feeding activity, composing the sound signals and the information of the head position for detecting the feeding activity and identifying and quantifying the feeding behavior based on the analysis of parameters indicating the sounds, the head position and movements.
  • it also comprises at least an embedded system (based on microprocessors, FPGA or ASIC) for processing the signals received from the NFC, extracting information about the typical group behavior of such animal based on the data of the position of the animals and identifying the abnormal behavior indicating that such estrus of such animal to compare the data of the group composition recently identified with the data of the typical group composition extracted.
  • such device in the invention also comprises preferably at least one embedded system (based on microprocessors, FPGA or ASIC) for processing the information generated by other blocks, extracting typical behavior information about such animal based on information generated by the blocks mentioned above and identifying abnormal behavior indicating estrus in such animal. It also comprises, at least a non-volatile storage system to store information generated by systems of processing and analysis of information implemented in the device. It also comprises, at least one transceiver unit to transmit the animal' s status and any other information obtained from the data processed, and to receive any other information obtained from data processed and to receive information from the remote transmitter, such as the operation parameters and when other animals have been identified as being in estrus cycle.
  • embedded system based on microprocessors, FPGA or ASIC
  • the invention device preferred mode also comprises, at least one source of power capable of obtaining solar energy from the use of solar panels, or any other alternative source of power, to store such power and to make an efficient use of it based on the use of high-performance chargeable and adjustable devices.
  • it preferably comprises a concurrent performance of tasks from the device to reduce its frequency of operation to the minimum level, and, therefore, with a reduction in the power consumption, which combined with a source of power based on claim 18 allows a continuous operation of the device for a continuous monitoring of the animals.
  • FIG. 1 Block diagram illustrating a possible implementation of the device in accordance with the invention, displaying its fundamental blocks
  • FIG. 1 Schematic illustration of the device location, with its sensors and antenna
  • Figure 3 Block diagram illustrating a possible implementation of the block for the analysis of the animal's feeding behavior
  • Figure 4. Illustration of the electrical outlet of the sound sensor unit of 30 seconds during a) grazing and b) rumination; Figure 5. Illustration of the different head positions of an animal: a) when it is in a group, b) when it is ruminating, and c) when it is feeding;
  • FIG. 1 Block diagram illustrating a possible implementation of the block for the analysis of the animal's feeding behavior
  • Figure 7 Block diagram illustrating a possible implementation of the block for the analysis of the animal's individual physical behavior
  • Figure 8 Illustration of the electrical outlet of the inertial measurement unit for an animal's movements during a) normal and b) estrus behavior
  • Figure 9 Block diagram illustrating a possible implementation of the classifier of individual physical activity and mounting detection
  • Figure 10 Block diagram illustrating a possible implementation of the block for the analysis of group behavior
  • FIG. 1 Illustration of the mounting process and communication between devices using a near-field communications network (NFC) ;
  • NFC near-field communications network
  • FIG. 12 Block diagram illustrating the implementation of the block for the analysis of the animal' s behavior, storage and communications
  • FIG. 13 Block diagram illustrating a possible implementation of the feeding block
  • FIG. 14 Block diagram illustrating the concurrent implementation of the device using four processors.
  • This invention provides a non-invasive device and a process comprising the stages of detection and analysis of signals indicating the physical activity (individual physical behavior) , feed (feeding behavior) and group behavior of an animal, identifying when deviations in the behaviors under analysis occur, as a whole, with respect to the normal behavior.
  • the identification of the feeding periods and their changes, added to the changes in the group behavior of each animal helps minimizing misinterpretations of the physical activity changes outside the normal behavior, in order to obtain more accurate and precise identifications of the state of estrus.
  • Such state is detected through the analysis of the animal's feeding, group, and physical activity behaviors using sorters based on computer intelligence tools (fuzzy logic, decision tree, neural networks, among others) or statistics.
  • This invention also provides a device comprising sensors for the detection of an animal's physical activity, the feeding activity and the group activity.
  • An embedded system based on microprocessors, programmable logic devices (FPGA) , or on application-specific integrated circuit (ASIC) that records and analyzes movement, position and sound data surveyed. Each one of such signals is processed independently for their conditioning and for extracting indicative information for each of the aspects of the animal's behavior under consideration.
  • the device may be fitted, at least partially, inside a watertight box that is placed preferably in accordance with the invention implementation, and by which the animal's head inclination and movement, the sound generated when feeding and its position may be detected. In other forms of the invention implementation, the case may be mounted on around the animal's neck using a belt.
  • An indicator showing when the animal is in estrus may be attached to the box.
  • the detection process may be completely performed in the embedded system and the notice may be sent when it has been identified that the animal is in estrus.
  • Another alternative consists in that only one part of the process may be performed in the embedded system and the information recorded and processed may be transmitted to a remote server to detect the animal' s status .
  • the analysis of the individual physical behavior of the animals may be based on statistical measures of the movement data recorded, as compared with a reference database. Another possibility may be the implementation of classifiers based on computer intelligence tools (neural networks, decision trees, etc.), which classified the data recorded, or signal characteristic parameters for the determination of the animal's status.
  • the preferred movement sensors are the inertial measurement units (IMU) based on MEMS and GPS-based technology.
  • the IMUs include an accelerometer, gyroscopes and magnetic compasses in a single device, which allow correcting, based on the signal processing, the sensor errors and settling the gravity effects. By analyzing and combining the IMU signals adjusted, information about the animal's movement may be extracted.
  • This information allows defining, for example, the animal's level of movements, the mounting moments or identifying any other type of actions that indicate estrus.
  • the physical activity level may be obtained from the energy or the root mean square of the movement signal adjusted at different time intervals.
  • the mounting moments may be identified from the detection of sudden changes and short duration of the vertical direction acceleration.
  • a recognition system may be used, combining signal processing tools and machine learning, which may operate in two stages. During the first stage, the IMU adjusted signal is converted into a set of predefined characteristics (frequency information, statistical attributes, energy, etc.) During the second stage, a classifier uses a set of characteristics and the outcome is the respective animal's status.
  • the sensors may be mounted on the animal' s neck or head for measuring its movement. If they are on the animal's neck, such sensors will amplify the detection work, as they only record the animal's body movement. While if the sensors are on the animal's head, they will measure the combined head and body movement. In the latter case, it will measure the animal's head inclination and movement with respect to the ground, allowing the detection of feeding periods and the improvement of the estimation of the feeding activity, as when the animal eats, its head tends to go down to the ground and move, while, generally when the animal is active and/or ruminating, the animal maintains its head higher and quiet.
  • estrus In estrus, the percentage of time that the animal' s neck is low (eating) is significantly shorter in comparison with the time in normal periods, as the animal consumes less pasture during estrus.
  • a combination of a high level of movement of the animal' s head and high values of head inclination may indicate a high probability that the animal is feeding.
  • the animals in estrus try to mount other animals. If the latter allow being mounted, it may be proved that the animal is in estrus.
  • the mounting process may be detected by two sensors: i) by movement sensors (accelerometers, inclinometers, GPS, etc.), which detect the jumping moment, and ii) by devices detecting the proximity of two animals, such as the near-field communication networks (NFC) and radio frequency identification device (RFID) . These two devices have in common that the communication distance between sender and receiver may be controlled, ensuring a maximum communication distance.
  • Another way of detecting the proximity of two animals is by sharing-via a communication network-and comparing the position of each of such animals as indicated by a GPS.
  • a high-level variation in the vertical acceleration signal needs to be identified, as the animal that mounts another animal has to jump and when it falls it hits the body of the animal that is underneath. Such sudden up and down movement is distinctively evidenced in the acceleration extent, as variations are noticed within a short period of time, which may not be identified when the animal is doing other activities. Also, a recognizer could be trained to detect such events from the acceleration signals measured.
  • the analysis of the animal's feeding behavior may be based on the analysis of the sounds that are normally produced when the animal is feeding. When the animal performs any feeding activity (grazing or ruminating) , its jaw generates sounds that may be recorded by a sensor, and then analyzed for determining the feeding activity performed.
  • the animal's head and neck are examples of adequate areas for the recording of sounds.
  • the preferred sound sensors are the electric directional microphones, which are capable of recording very weak sounds in the preferred direction (front of microphone), however, any other type of microphone (for example, MEMS) may be used.
  • MEMS magnetic resonance sensor
  • Two configurations may be used: i) Just one microphone if the acoustic insulation is good, such that the signal recorded has a good relation signal / sound, or ii) (two microphones) to be able to make an active cancellation of the ambient sound in more unfavorable conditions.
  • the signal is previously processed by a cleaning algorithm that mitigates the ambient sound effects and other undesired signals.
  • the resulting sound signal is analyzed to determine when the animal is eating or is ruminating.
  • the sounds produced during the ingestion of food are more intense and shorter than when the animal is ruminating, which are lower and regular, with different repetition frequencies.
  • the reason for such difference lies in the components acting in each of the mechanisms, in addition to the consistency and the contents of water which the matter processed has and its availability.
  • signal processing tools and computer intelligence it may be determined when the animal is feeding or ruminating.
  • the parameters detected may be aggregated, and using statistical processes or computer intelligence tools, the reference parameters may be determined for characterizing normal feeding activities.
  • the animal's feeding activities may also be additionally evaluated by analyzing the head position with respect to the ground and its movements.
  • the signals created by the accelerometers include head position information, from the head weight direction (gravity force) , as well as the movements made. A short distance between the head and the ground, added to a movement increase may indicate that the animal is eating. The distance may be measured using accelerometers and gyroscopes based on MEMS technology. This combination of sensors allows measuring both, the head inclination angle and its movement. So, it may be expected that, when the neck inclination sensor measures a large angle (the animal's head near the ground) and the intensity of the head movements is high, the animal may be feeding.
  • the animal may be ruminating. So, the statistical tracking of the signals of such sensors may be used to evaluate the animal' s reproductive status.
  • the analysis of the group behavior of the animals may be based on the use of clustering techniques. Such techniques allow finding the groups of animals with similar behaviors in addition to other data, such as the relative mean distances among the animals along the time, providing an estimation of candidate groups to be in estrus and feasible of being studied using supplementary information.
  • clustering techniques allow finding the groups of animals with similar behaviors in addition to other data, such as the relative mean distances among the animals along the time, providing an estimation of candidate groups to be in estrus and feasible of being studied using supplementary information.
  • Such behavior may be monitored through a device to determine which animals are close to the estrus cycle, comparing the information obtained with a database with information about the animals in the estrus cycle in the cattle. Then, we have made a statistical analysis of the composition of the group of animals, comparing with the information obtained from a database on the status of the animals.
  • the database is periodically updated to follow up the changes of status of the animals along the time.
  • Each animal in the cattle is identified through a single identification code, which is stored in the devices.
  • a form of detecting the proximity of the animals is by using NFC networks. Such devices allow exchanging information controlling the distance of communication between the sender and receiver. Thus, the animals that are at a certain distance from a given animal may be explored.
  • Another form of detecting the proximity between animals is by sharing -via the communication network- the code of identification and the position of the animals. Once detected the animals that are close to the animal under study, such information is compared against a database on the animals in estrus at that moment. From the statistical analysis of such information, the probability animal under study is in estrus may be determined with the group behavior information of such animal. Also, this block allows detecting when a specific male mounts a cow and the genetics of the cattle may be followed up.
  • the information corresponding to each of the behaviors analyzed for each animal are integrated along the time and with any other relevant information (estimated date of estrus, milk production, etc.), to determine if an animal is in estrus or not.
  • the analysis may be based on the use of machine learning (support vector machines, neural networks, decision trees, etc.) or statistical analysis that allow determining when the animal's behavior deviates from the normal behavior and gets closer to the estrus behavior.
  • machine learning support vector machines, neural networks, decision trees, etc.
  • statistical analysis allow determining when the animal's behavior deviates from the normal behavior and gets closer to the estrus behavior.
  • the use of such sources of information allows distinguishing such changes in just one of the aspects of the animal's behavior, decreasing the number of false alarms produced by feeding problems, changes of diets and any other anomalous situation that may affect the characterization of a given animal. Once determined the status of the animal, it is reported to the system.
  • This invention uses the indicative information of the physical activity (movements of the animal) in combination with indicative information of the feeding activity (sounds and head position) and information indicating its group behavior (group structure) to determine if an animal is in estrus or not.
  • the integration of such three sources of information allows increasing the accuracy and correctness of the detection of the estrus to properly interpret the animal' s behavior changes .
  • This invention includes a device comprising sensors to record the animal's movements, its position, the sounds produced during feeding and identifying the animals close to it.
  • An embedded system based on microprocessors, programmable logic devices (FPGA) , or an application-specific integrated circuit (ASIC) that records, processes, and analyzes all such signals independently to extract information that indicates each of such aspects of the animal's behavior.
  • the estrus detection process may be performed as a whole in the embedded system and the resulting information is periodically sent to a computer.
  • Another alternative is that only a part of the process is performed in the device (such as the recording and conditioning) and the information obtained is transmitted to a computer for its analysis.
  • the device may be partially fitted into a watertight box that will be preferably located on the animal ' s head or neck, in accordance with the invention implementation.
  • a watertight box that will be preferably located on the animal ' s head or neck, in accordance with the invention implementation.
  • Such locations allow easily detecting the animal's movements and position, and also when it is feeding (rumination or grazing) some sounds are generated that may be surveyed by means of a sensor on the animal's head and then analyzed to determine if the animal is in estrus or not.
  • the preferred sound sensors are the directional Electret microphones, which are capable of recording very weak sounds in the preferred direction and its response in the frequency of interest band is flat. As the signal is weak, it must be fitted (amplified and filtered) through specific analogical electronic circuits and, then, digitize it.
  • the digitized audio signal is previously processed by a cleaning algorithm that mitigates the ambient sound effects and other undesired signals.
  • the preferred movement sensors are the inertial measurement units (IMU), which are capable of recording translational accelerations, angular speeds and the position with respect to the Earth's magnetic north. Such signals are processed by estimation and filter algorithms to remove the effects of noises and errors (offsets and bias) .
  • the preferred position sensor is a GPS, which may include any correction signal to improve the accuracy of the data measured.
  • Figure 1 presents a block diagram of an estrus detection device built and operating in accordance with the implementations of this invention. Six functional blocks are included:
  • this block records the sounds and movements of the animal's head when feeding to determine and quantifying its feeding activity.
  • the signals recorded are processed and suited for removal of any noises and disturbances that may have been recorded together with the sounds made by the animal when feeding.
  • the head movement signal is specifically processed to remove systematic errors from sensors and any noise. Then, the sounds are analyzed and added to the head movements and position for detecting and quantifying the feeding activities (grazing or rumination) performed by the animal in real time.
  • this block records the animal' s movements along the time to determine and quantifying the animal's physical activity.
  • the signals recorded are the translational accelerations, the rotation speeds and the orientation of the animal's head with respect to the magnetic north. Then, such signals are processed and suited for removal of any systematic errors and noises that may have been recorded with them.
  • the animal's movement signal is analyzed for detecting and quantifying the physical activity level in real time, as developed by the animal and for detecting the mounting time by the animal.
  • the analysis of the movement signal also allows obtaining a history of the physical activities developed by the animal, enabling a comparison of histories of different days for detecting any changes in the animal's behavior.
  • this block records the composition and structure of the group to which the animal belongs along the time.
  • the signals recorded are the animal' s position in the field and the animals that are near this animal.
  • Such signals are processed and suited for removal of any errors and noises that may have been recorded with them, and are analyzed for detecting and quantifying in real time the number of animals in estrus that are in the group of the animal being monitored.
  • this block integrates and analyzes the information produced by the previous blocks with any other information available for detecting if the animal is in estrus or not.
  • this block organizes all the information produced by the device from the information recorded. It organizes the information in self-contained blocks that provide additional useful information (such as, the time and place where such data was recorded, among others) . Also, it exchanges the information with the system through a wireless communication system. This allows communicating the animal ' s information to the management system in real time.
  • this block manages the power and availability of the device batteries to maximize its operation time. This task is performed by monitoring the battery charge, the power provided by the power collection system and by disabling the device functions, based on its priority .
  • Figure 3 illustrates the implementation of the feeding activity analysis module.
  • the sound signal recorded by the microphone follows two different routes: i) One for detecting and sorting out the events, and ii) the other for calculating the characteristic parameters of the activity.
  • Each of such routes receive different adjustments before their processing.
  • the signal is filtered and amplified for reducing the effects of any noise and disturbances, and, then, the signal is digitized by an analogical/digital converter.
  • the signal used for detecting the events is suited with an amplifier with an automatic gain control (AGC) to obtain a signal with a good signal/noise ratio and without any disturbances. This improves the recognition and sorting rate, in connection with cases where processed with a fixed gain amplifier.
  • AGC automatic gain control
  • the signal used for calculating the parameters is suited by a fixed and known gain amplifier, such that the relation between the original signal and the processed signal may be determined and, thus, the power contained in the original signal may be calculated.
  • a directional Electrect microphone is used, which is fastened to the animal's forehead, as in U2 ( Figure 2) .
  • the animal' s head movement is recorded through an inertial measurement unit (IMU), which records the translational accelerations and the angular speeds related to the head movement.
  • IMU inertial measurement unit
  • the angular position may be determined with the accelerometers from the estimation of the gravity vector and the use of trigonometrical functions. However, the accelerometers measure all the forces acting against the object, for which measurements may be contaminated with short disturbances, making the data obtained from the accelerometers reliable only in the long run.
  • the angular position may also be determined with the gyroscopes, which measure the angular speeds.
  • such sensors have drifts (they do not return to zero when the object returns to its original position) , which introduce systematic errors in the long run.
  • the estimation block calculates the head movements and position.
  • an IMU is used, located at the Ul device ( Figure 2) .
  • Figure 4 shows the microphone outlet during 30 s during grazing ( Figure 4. a) and rumination ( Figure 4.b) .
  • the sounds produced during grazing are characterized for being short, without a very defined rhythm (but approximately of 1 Hz) .
  • the sound produced during rumination lower relative energy and a more noticeable periodicity, at two levels: the level of rumination during rumination (with a frequency of about 1 Hz) and the level of regurgitations (repeated approximately every one minute) .
  • the animal may be standing still or lying during rumination, and there is an average of 15 to 20 periods a day where the animal regurgitates from 300 to 400 portions of food, with an average of 50 chewing movements per portion.
  • the first step to analyze the feeding activity is to identify and characterize chewing events: bite, bite-chewing, and chewing.
  • Each such actions produce sounds, whose frequency contents will be given by the material processed, while the duration and amplitude of the sound will be determined by the movement of the jaw.
  • the sound produced by a bite is characterized by a high amplitude and a short duration, which is characterized of a tear.
  • the sound produced by chewing is characterized by a small amplitude and a longer duration, which is characterized of a squeal.
  • the combined movement of chewing-bite is the combination of both.
  • a noise means a sound that does not satisfy any of the characteristics described above and a silence is the absence of sounds during the period of an average event.
  • the animal's feeding behavior may also be evaluated through the head position with respect to the ground and its movements.
  • a short distance (large angle) and an increase in the number of small-amplitude local movements may indicate that the animal is eating ( Figure 5.c), while a long distance (small angle) and a few local movements may indicate that the animal is ruminating (Figure 5.b) .
  • the IMU accelerometers are used for measuring the head position and movements. Such sensors allow measuring both, the inclination angle and the head movements.
  • Figure 5 shows illustrations of some head positions of an animal in different situations. When an animal is not eating, the inclination may be as illustrated in Figure 5. a and 5.b. When the animal eats food, the inclination increases, as the head goes down the ground, as illustrated in Figure 5.c.
  • the head inclination When an animal is ruminating, the head inclination may be as illustrated in Figure 5.b.
  • a popular method for the interpretation of head inclination data is the use of (fixed or adaptive) thresholds in the head inclination angle. When the inclination signal recorded exceeds this threshold, it means that the movement signal must be interpreted as the representation of a feeding period.
  • the threshold used for sorting out may be fixed or automatically updated with the information gathered from a given time window.
  • the chewing events occurred during the feeding activity may be identified from the analysis of the most relevant temporary characteristics of the sounds produced by the animal when feeding (shape of the power envelope, maximum amplitude, duration, periodicity, sequence and symmetry, among others) , combined with the analysis of the head movement patterns and its angle with respect to the floor. Then, the temporary characteristics are processed by a sorter together with the patterns of movements and head angles for the event identification.
  • the events identified are: bite, chewing-bite, chewing, and silence.
  • Figure 6 shows the implementation of the analysis block for feeding activities, which is built by three sub-blocks: i) an extractor of characteristics, ii) a sorter and calculation of the parameters of chewing events, iii) a sorter and calculation of parameters of feeding activities.
  • the extractor of characteristics processes the acoustic signal emitted by the AGC, and the signal of the head position and movements to extract the most relevant temporary characteristics of the signal.
  • the power envelope of the audio signal recorded, which is sub-sampled.
  • four temporary characteristics are calculated: maximum amplitude, duration, shape (sign of the derivative) and symmetry. But others may be calculated: The maximum amplitude is directly detected from the sound signal, in a temporary window, with the size of a chewing-bite event, as the event is longer. The remaining characteristics are determined from the signal power envelope.
  • Such characteristics of sound signals and the head movements and position are analyzed together by the event sorter.
  • This block determines if a chewing event has occurred or not, and what type of event (chewing, chewing-bite, bite) has occurred.
  • some computer intelligence tools may be used, such as the neural networks, or the decision trees, among others, which are trained for such purpose.
  • a multi-layer perceptron is used for providing the event tag.
  • the event parameter calculation block uses the event tag information combined with the digitized signal without AGC for calculating the power of the events. The remaining characteristic parameters of the events (duration, number, maximum amplitude, etc.) are calculated based on the characteristics generated by the sub-block of extraction of characteristics.
  • the activities carried out by the animal are identified and quantified for the period under consideration, by means of the feeding activity sorter.
  • the information about the events (tag and parameters) is used by the sorter of activities, which implements a statistical model based on decision trees for the evaluation of the sequence of events identified ( Figure 6) .
  • the activities identified are: grazing, rumination, and none. For example, if a sequence of rumination events is identified, with any silence or noise alternated between them, it is highly probable that the animal was ruminating all the time, so that the tags for the alternated events are adjusted.
  • the parameters that characterize the group of events are calculated (rumination time, grazing time, number of events, grazing power, among others) and, then, the amount of dry matter ingested is determined from the consumption model.
  • Figure 7 illustrates the implementation of the analysis of the individual physical behavior.
  • the module is composed of an inertial measurement unit (IMU) that measures the animal's head movements and a GPS that measures the animal's absolute position.
  • IMU inertial measurement unit
  • sensors are located at the animal's head in Ul ( Figure 2) or on its neck.
  • IMU inertial measurement unit
  • Such sensors have systematic errors (bias, drifts, etc.) and significant random errors that make more difficult the proper estimation of the movements and position, for which the information from such sensors must be combined for error correction.
  • the algorithm used for obtaining the signals of movement and position adjusted is a regressive least-squares estimator. Such adjusted signals may be directly analyzed or used as input for another processing block.
  • Figure 8 presents the measurements of the acceleration module during a normal activity period ( Figure 8. a) and a period in which the animal is in estrus ( Figure 8.b) . As it may be noticed in Figure 8. a, the acceleration during normal activity has less intensity.
  • the acceleration during estrus tends to be more intense, with very high peaks.
  • Figure 9 shows the implementation of the individual physical behavior analysis block, which is built by five sub-blocks: i) estimation and filtering, ii) segmentation, iii) extraction of characteristics, iv) mounting detection, and v) physical activity classification.
  • the estimation and filtering block estimates the body position, its movements and the head movements from the information provided by the IMU and the GPS. From the regressive least-squares estimation algorithm, this block estimates the information used by the detection and sorting algorithms (body position, body movements and head movements), minimizing the effect of noises and systematic errors.
  • the body movements are represented, together with the head movements, in the adjusted acceleration signal resulting from the estimation and filtering block.
  • the animal' s body movements may be estimated along the time. When there are GPS measurements, this information may be adjusted.
  • the segmentation block divides the acceleration signal into segments of a few seconds for delivery of the signal to be processed to the extraction block. This temporary division of the signal allows having an almost continuous description of the physical activity developed by the animal, as the analysis of each segment of the movement signal allows the identification of the physical activity developed in such interval.
  • the characteristic extraction block processes the segmented signal for the extraction of temporary and frequency attributes that gather discriminative information from it.
  • the activity classifier block is composed of a static classifier (such as a decision tree, a neural network, or a support vector machine) that is trained to recognize the activity to which each characteristic vector corresponds, which may be sampled at their input.
  • a static classifier such as a decision tree, a neural network, or a support vector machine
  • the classifier in its training phase, builds models for each physical activity to be recognized, discovering similarities between vectors of characteristics of the same activity, and establishing a general relation between characteristics and activities. Such relation is fixed after the completion of the training stage and is applied every time the system presents a characteristic vector to this block.
  • Mounting between animals may be detected by analyzing the vertical component of the acceleration signal, as it shows the ascending and descending movement that is characteristic of such action ( Figure 11) .
  • This process may establish a threshold that may only be exceeded when the amplitude of the variation of the vertical acceleration is caused by a jump when the animal mounts another and then analyzes the characteristic of a period of the signal focused on the possible event. This analysis will be focused on the morphology of the vertical acceleration signal power envelope.
  • the movement signal characteristics, the animal's position and the occurrence or not of a mounting event are analyzed together by the activity classifier. This block determines the activity type being performed by the animal and if its parameters are deviated from the expected normal behavior.
  • some computer intelligence tools may be used, such as the neural networks, or the decision trees, among others, which are trained for such purpose.
  • a multi-layer neural network is used for providing the tag for the activity performed. The activities identified are: Standstill, Lying, Walking, Running and Being Quiet. Once completed the data block activity type, the characteristic parameters of the group of events are calculated.
  • a direct analysis of such signals may be performed to obtain, for example, the history of how the animal ' s activity level varied or how many times it mounted other animals or was mounted by others and when that happened.
  • the activity history may be obtained when calculating the power or the average standard deviation at fixed intervals of time of the acceleration signal during such interval.
  • the activity analysis module provides relevant information to other system modules:
  • the universal time is obtained from the GPS information, while the head movements and mounting events are obtained from IMU measurements.
  • Figure 10 illustrates the implementation of the analysis of the group behavior.
  • This module is composed of a network of near-field communications that records and identifies the animals within a certain distance (approximately three meters) away from the animal being monitored, starting from the transmission antennas placed in the muzzle in U3 ( Figure 2) and the reception antenna located at the device, in Ul ( Figure 10).
  • the time measured by the behavior analysis block is used for a temporary tracking of the group composition and to define a database describing the group composition.
  • Such real ⁇ time database is compared against a database of animals in estrus within the cattle, which is periodically updated. Such comparison allows measuring the group composition.
  • This group may be characterized from statistical measures of the group representing the group relevant information and/or of computer intelligence tools. From the use of classifier based on mechanical learning tools (such as, fuzzy logic, decision trees, etc.), the group composition information and a database of the females in estrus and the males available, a reproduction behavior of the group to which the animal being monitored belongs may be sorted out. Such information is sent to the animal's status analysis module. In case a mounting attempt is detected, the device communicates with the closest cow, which is supposed to be the mounted cow, to exchange the animal's identification codes and update the information about the mounting attempt in both animals involved ( Figure 11) : the mounting animal and the animal that was mounted.
  • Figure 12 illustrates the building of the behavior, storage and communications analysis block.
  • This block determines the animal's status and organizes the information generated by the three analysis blocks, and other sensors, for its storage and transmission.
  • This block receives the feeding behavior information, individual physical and group behavior, where it is processed to determine the animal's reproductive status.
  • the block uses computer intelligence tools that let it integrate the independent information generated by each module and obtain a more accurate result.
  • this block organizes the information in a self-contained data package, to which a detection/adjustment code is included without requiring the re-transmission of information.
  • This package includes the geographic position where the data was obtained, adding a traceability element to the information generated by the device.
  • the information package is stored in a non-volatile memory and eventually transmitted to a personal computer for its later processing. The transmission of information will be performed through a low-power Wi-Fi interface. Such communications will be periodical by day and the information generated by night will be stored and transmitted at a predefined time to save power.
  • Figure 13 illustrates the building of the power management block.
  • Such block generates the different power source lines (analogical and digital) used by the recorder from a single source of power (batteries) and manages the power available to maximize the recorder operation time.
  • Such task is completed in two supplementary ways: i) By ensuring the maximum possible battery charge from the power provided by a solar panel, or any other power gathering device, and ii) controlling the recorder functions that must be operative every time.
  • This block has a high-efficiency DC-DC regulator combined with low- dropout regulators to maximize the power efficiency and enable the recorder operation, even in cases where the battery voltage is lower than the recorder operation voltage.
  • the sensing unit comprises a microphone for detecting the sounds produced, an inertial measurement unit (IMU) for measuring the animal's head movements, a GPS for determining the animal' s position, a transceiver for near-field communications (NFC) with its antennas for determining the animals that are near and a Wi-Fi transceiver.
  • Sensors, antennas, and the processing unit are located at the animal's head and a muzzle is used for their fastening, as indicated in Figure 2.
  • the microphone used may be any type of sound sensor known with directional features (it only reads the sounds coming from the front area) and is located on the inner side of an adjustable band (which may be an elastic band) of the muzzle, which is placed on the upper part of the animal's head, as shown in U2.
  • the IMU, GPS and NFC transceiver are located together with the Ul processing unit located at the back of the head, on the neck ( Figure 2) . Such location allows maintaining short connections with the microphone and the antennas, protecting the recording unit and the animal promptly naturalizes its presence.
  • the adjustable band is protected by a waterproof cover and a foam rubber or similar material to protect the microphone against ambient conditions. The combination of the directionality of the microphone catchment area, combined with the passive insulation of ambient noises of the materials covering the band are sufficient for operating without any inconveniences under the field operating conditions frequently prevailing.
  • the NFC transceiver transmission antennas are located on the muzzle in U3 ( Figure 2), while the receiving antenna is in the device, in Ul . Thus, the transmission and reception antennas are at a distance of between one and two meters when an animal is mounted.
  • the processing unit is composed of four processors and a set of specialized integrated electronic circuits used to implement the elemental units in the processing unit ( Figure 14) : i) signal acquisition and processing unit, ii) the unit of analysis of activities and behavior, Hi) the storage and communication unit, and iv) the power supply and management unit.
  • the signal acquisition and processing unit is implemented through three electronic circuits dedicated to each of the signals recorded: sound, movement, and position.
  • the sound and head movement signals are analyzed by processor 1, which analyzes the animal's feeding behavior.
  • the movement signals are processed by processor 2, which adjusts systematic errors and removes the sounds of the movement signals.
  • the animal's movements and signals are analyzed by processor 3, which analyzes the individual physical behavior and the animal's group behavior.
  • processor 4 determines the animal's status (combining the information of each aspect of the animal's behavior), builds data packages, and transmits such packages and communicates with other devices. As his processor performs the tasks that are more power-consuming (wireless communication) , it also deals with the device power management.
  • This implementation concurrently with the tasks performed in the device, allow reducing the clock frequency of the processors at very low values, due to the concurrent and parallel performance of tasks, so that the ultra-low power operation modes of commercial processors may be used.
  • this implementation of the device allows its operation at a very low power, enabling continuous operation of the device using solar energy or any other alternative source of power.
  • the sound produced by the animals is recorded through a directional Electrect microphone connected to two channels from an integrated acquisition system ( Figure 2) .
  • the acquisition system has been implemented in a low-power integrated electronic circuit including in each of its two channels: a programmable gain amplifier, a bias circuit for microphones, an automatic gain control adjustable for the noise level and programmable parameters and an analogical- digital 24-bit converter and a programmable sampling frequency of up to 96 KHz.
  • the integrated circuit includes a digital SPI serial interface dedicated to programming of operation parameters and an audio I2S digital serial interface for reading data. This implementation was used at a sample frequency of 2 KHz, with microphone bias circuits disconnected and the automatic gain control enabled only on the right channel.
  • the left-channel amplifier gain was adjusted to a fixed value of 12 dB ( Figure 3) .
  • the resulting data was read by the processor by means of an I2S interface implemented by a dedicated routine. Programming of the acquisition circuit parameters was performed by means of an SPI interface, implemented through dedicated hardware available in the processor. The signals obtained are re-quantified at 8 bits to use fixed point arithmetic in the algorithms.
  • the animal's head and body movements are recorded through an IMU.
  • the acquisition system has been implemented in the same low-power integrated circuit, including three accelerometers , three gyroscopes, three magnetometers of 16-bit resolution and a bandwidth of 100 Hz, a data fusion unit and a power management unit.
  • the integrated circuit includes a digital I2C serial interface dedicated to programming of operation parameters and reading of data.
  • the circuit reads the data from the sensors and merges them for correcting any errors and removal of noises, using the least-squares estimation algorithm (Kalman filter) .
  • the raw data provided by the sensors and/or characteristic parameters of the movement calculated by the merge unit may be read by the user, such as the Euler angles, the gravity vector and the linear accelerations, among others.
  • This implementation used an estimation algorithm combining measurements from three sensors
  • the sampling frequency of the IMU was set to 10 Hz., with sensor ranges limited to ⁇ 2 g/s for accelerometers and ⁇ 250 °/s for gyroscopes, to obtain the maximum sensitivity. From all the information available, the following may be read: i) Euler angles, ii) the gravity vector, and Hi) the linear accelerations through the processor dedicated I2C interface.
  • the unit for analysis of feeding activities is implemented in processor 1 through the routines used for processing and analyzing the data obtained from the microphone and the IMU. These routines implement in the microcontroller the algorithms for the detection and classification of events and activities
  • Figure 6 in addition to the routines required for extracting, classifying, and quantifying the information indicating the individual events.
  • routines process the signals recorded as the measurements are recorded using the fixed-point arithmetic.
  • the right-channel audio signal power envelope is calculated (as processed by the amplifier with automatic gain control) from the filtering of the signal absolute value.
  • the signal absolute value is calculated by setting the sign of all data, while filtering is performed using an infinite impulse response (IIR) filter of second degree implemented using the fixed-point notation with 8 bits of resolution.
  • IIR infinite impulse response
  • the information of the signal power envelope from the sound block is sub-sampled at 100 Hz ( Figure 6) .
  • the sub-sampling process is implemented by software on the data recording of the signal power envelope stored in the micro-controller memory. Once obtained the sub-sampled audio signal power envelope, the signal characteristics are calculated for sorting out purposes, the characteristics used are: shape, duration, symmetry and maximum window value.
  • the signal power envelope shape is determined from the signs of the slopes and the duration is obtained by counting the number of samples that are higher than a predefined value.
  • the gravity vector and the Euler angles calculated by the IMU are used for supplementing the audio information.
  • the gravity vector is used for determining the distance from the head to the ground, from the head angle with respect to the gravity acceleration vector, while the Euler angles are used for characterizing the head movements.
  • These signals are used for identifying the feeding activity performed by the animal and the accelerations are used for the individual behavior analysis (the physical activity level) .
  • the algorithm generates temporary flags indicating the presence of potential events and the signals are segmented . Temporary signals are generated from a comparison of the signal power envelope with a threshold that varies along the time.
  • the threshold value is lower than the signal power envelope
  • such envelope is segmented taking 60 samples before and after the moment when the event was detected.
  • head angle gravitation vector
  • Euler angles the temporary characteristics of the signal power envelope
  • shape, duration, symmetry and maximum value are used for detecting and sorting out events ( Figure 6) .
  • the event sorter is implemented as a multi-layer perceptron sorting out acoustic events detected in: Bite, Chewing, Chewing-Bite, Silence and Noise.
  • the characteristic parameters are calculated for each individual event: event duration and power.
  • Each event power is calculated from the signal power envelope of the audio left channel (as processed by the fixed- gain amplifier) , which keeps the power of the signal recorded. Only the chewing power is calculated. This information, combined with the number of bites and their duration, is used by the consumption model for estimating the number of dry matter ingested by the animal in the period of time under consideration ( Figure 6) .
  • Another routine takes the results from the event sorter and they are organized in packages.
  • a package is defined as a set of events in a 15- minute period. Then, once the packages have been built, basic statics are calculated (number of events, types of events, duration and power, among others) and the feeding activity developed is analyzed and quantified using the fuzzy logic.
  • the unit for analysis of individual activity is implemented in processor 3 through the routines used for processing and analyzing the data obtained from the accelerations measured by the IMU.
  • These routines implement the algorithms for the detection and classifying mounting activities and detection ( Figure 9) , in addition to the routines required for extracting, classifying, and quantifying the information indicating the individual physical activity.
  • Such routines process the signals recorded as the measurements are recorded using the fixed-point arithmetic.
  • the acceleration signal is processed in parallel for i) determining and quantifying the physical activity developed by the animal and ii) detecting mounting events.
  • the tasks required for determining the physical activity involve the segmentation of the acceleration signal in blocks of five seconds, at which the most relevant characteristics are calculated, which allow characterizing the physical activities carried out by the animal.
  • temporary attributes are statistical measures of the acceleration signal (mean value, variance, kurtosis, power, amplitude, maximum, minimum) but others may be calculated.
  • the frequency attributes are the coefficients obtained by the cosine transform, which has computing advantages and is equivalent to the discrete transform of Fourier, in this case.
  • the frequency attributes may also be produced by the use of filter banks providing information by means of predefined frequency bands. All such information, combined with the detection of mounting events, is used for detecting and sorting out the physical activity carried out by the animal during analyzed period ( Figure 9) .
  • the event classifier is implemented as a multi-layer perceptron classifying the activity detected in: Standstill, Lying, Walking, Running and Quiet. Also, the characteristic parameters are calculated for each activity: event duration and power.
  • Each event power is calculated from the acceleration module. Once detected, classified, and quantified the activity of the segments, another routine takes the results from the activity classifier and they are organized in packages. A package is defined as a set of activities in a 15- minute period. Then, once the packages have been built, some statistics are calculated.
  • processor 3 communicates with the other animal by means of an NFC transceiver ( Figure 10), which allows reading the identification code of the animal mounted and transfer its own code to another animal ( Figure 11) . If the mounted animal agrees being mounted, the mounting animal will not move, so that vertical acceleration of such animal is very little (almost null), with which, mounting will be confirmed and the mounted animal's identification code will be stored. Such code is transmitted to the system, together with the animal's information, and it is added to the information about the rest of the cattle.
  • the animals that are close to such animal are determined by using an NFC transceiver ( Figure 10) .
  • NFC transceiver Figure 10
  • a specialized integrated circuit has been implemented that includes the automatic management of communications, at physical layer and protocol level, of ISO/IEC 7816, ISO/IEC 14443 and EMVCo 4.3 standards, among others.
  • the device includes the radio-frequency interface, clock generation, automatic control of operation voltages and protection of the radio-frequency interface.
  • the communications system uses the ISO/IEC 14443 standard with the NFCIP-1 protocol, in active mode, with a transfer speed of 424 Kbit/s .
  • the group behavior analysis unit is also implemented in processor 3, as information similar to the physical behavior is used, through the routines used for processing and analyzing the information about the animal's position relative to the group.
  • routines implement the algorithms for detecting animals that are close to the animal being monitored, using an NFC transceiver ( Figure 10) .
  • Such transceiver allows communicating with another transceiver at a distance shorter than 3 m.
  • the tasks required for the determination of the group behavior involve detecting and identifying the animals that are close (less than 3 m away) from the animal being monitored and their status.
  • the animals that are close are identified by exploring their environment with the NFC transceiver, which establishes a communication with the devices that are close to it, obtaining an identification code from the other animals from the active communication between the transceivers, using the NFCIP-1 protocol of the ISO/IEC 14443 standard.
  • the status is determined after a search on a list of animals in estrus provided by the system through the communication system. Then, using a fuzzy logic-based system, the possibility that the animal is in estrus is determined.
  • the reproductive behavior analysis unit is implemented in processor 4 through the routines used for merging, fusing and analyzing the information generated by the individual and group feeding behavior analysis units, respectively ( Figure 12) . It communicates with other processors through a dedicated high-speed SPI interface, which allows transferring the information required for the execution of the algorithm, while maintaining the PCB plate design simple. Such routines implement in processor 4 the algorithms for detecting and classifying the animal's reproductive status. Such routines process the information provided by other processors, using an expert system based on a fuzzy logic. A report is generated periodically, which is stored in the non-volatile memory and sent to the base station by means of a low-power Wi-Fi transceiver ( Figure 12) .
  • the non-volatile memory of the device is built with at least 16 Gigabytes of low-power flash memory.
  • the communication module may implement point-to-point connections with a base station or other devices for implementing networks with star or mesh topology, as required, with transmission speeds of up to 250 kbit/s and a 500-meter range at open field, implementing the IEEE 802.15.4 standard. Both, factors such as module shape and consumption are little and support the device autonomy requirements.
  • the processor communicates with the non-volatile memory by means of an SPI interface, implemented through dedicated hardware available in the processor.
  • the communication module communicates with the processor through a UART asynchronous serial interface implemented through the dedicated hardware available in the processor.
  • the power source block generates the different power source lines used by the device from a single source of power and manages the power available to maximize the recorder operation time (Figure 13) .
  • This task is carried out in two complementary manners: i) ensuring the battery maximum charge possible from the power provided by a solar panel, and ii) controlling the recorder operating functions at all times.
  • the recorder functions are controlled by disabling the recorder functionalities, based on its priorities. In other words, to the extent that the battery charge decreases, the frequency of the use of wireless communications decreases, until its cancellation in extreme cases, for the purpose of ensuring the recording and analysis of the animal's activities.
  • This block manages the operation of the modules, a battery charger using solar energy and a high-efficiency DC-DC regulator combined with low-dropout regulators (Figure 13) , to maximize the power efficiency and enable the recorder operation, even in cases where the battery voltage is lower than the recorder operation voltage .

Landscapes

  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Veterinary Medicine (AREA)
  • Animal Husbandry (AREA)
  • Biophysics (AREA)
  • Environmental Sciences (AREA)
  • Zoology (AREA)
  • Wood Science & Technology (AREA)
  • Engineering & Computer Science (AREA)
  • Animal Behavior & Ethology (AREA)
  • General Health & Medical Sciences (AREA)
  • Public Health (AREA)
  • Biodiversity & Conservation Biology (AREA)
  • Pregnancy & Childbirth (AREA)
  • Fodder In General (AREA)
  • Feed For Specific Animals (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)

Abstract

L'invention concerne un procédé destiné à la détection de chaleurs chez un ruminant, fondé sur la détection et la quantification du comportement physique individuel de l'animal, du comportement alimentaire et du comportement en groupe, et un dispositif (U1, U2, U3) permettant la mise en œuvre d'un tel procédé.
PCT/IB2017/057967 2016-12-16 2017-12-14 Procédé et dispositif de détection de chaleurs chez un ruminant WO2018109725A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
ARP160103881A AR112047A1 (es) 2016-12-16 2016-12-16 Proceso para detectar celo en un animal rumiante y dispositivo
ARP20160103881 2016-12-16

Publications (1)

Publication Number Publication Date
WO2018109725A1 true WO2018109725A1 (fr) 2018-06-21

Family

ID=61005856

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/IB2017/057967 WO2018109725A1 (fr) 2016-12-16 2017-12-14 Procédé et dispositif de détection de chaleurs chez un ruminant

Country Status (2)

Country Link
AR (1) AR112047A1 (fr)
WO (1) WO2018109725A1 (fr)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109122371A (zh) * 2018-08-23 2019-01-04 贵州谷丰生态农业发展有限责任公司 一种肉牛放养用定位装置
CN111870387A (zh) * 2020-08-04 2020-11-03 青岛农业大学 多功能母猪发情检测装置
WO2021050775A1 (fr) * 2019-09-12 2021-03-18 Performance Livestock Analytics, Inc. Collecte et traitement de données de bétail et de parc d'engraissement à l'aide d'une interrogation de bande uhf d'étiquettes d'identification radiofréquence d'arrivée en parc d'engraissement et d'évaluation de risque
US20210295951A1 (en) * 2020-03-20 2021-09-23 Kpn Innovations, Llc Artificial intelligence methods and systems for generating zoological instruction sets from biological extractions
RU2757683C1 (ru) * 2020-09-22 2021-10-20 Общество с ограниченной ответственностью "Научно-производственная компания "Биосенсорика" Способ определения состояния половой охоты сельскохозяйственного животного
WO2021234490A3 (fr) * 2020-05-19 2022-01-06 Agverse Technologies Private Limited Système et procédé de gestion de nutrition et de détection de chaleurs et d'insémination chez des animaux producteurs de lait
WO2022115916A1 (fr) * 2020-12-04 2022-06-09 Finchain.Ai Pty Ltd Surveillance et gestion de bétail
WO2022154661A1 (fr) * 2021-01-18 2022-07-21 Nedap N.V. Procédé et système de détection d'œstrus chez un mammifère
US11617352B2 (en) 2018-01-23 2023-04-04 William R. Jackson, III Method and apparatus for detection of estrus and optimal time for embryo transfer or artificial insemination in animals

Citations (22)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US708375A (en) 1902-03-15 1902-09-02 Theodore Wick Lowe Windmill.
US4247758A (en) 1979-11-15 1981-01-27 Rodrian James A Animal identification and estrus detection system
US4618861A (en) 1985-03-20 1986-10-21 Cornell Research Foundation, Inc. Passive activity monitor for livestock
US4846106A (en) 1985-06-06 1989-07-11 Cowtronics, Inc. Method and apparatus for detecting standing heat in cattle
US5111799A (en) 1990-03-28 1992-05-12 Washington State University Research Foundation, Inc. Estrous detection systems
US5542431A (en) 1993-06-30 1996-08-06 Ddx Incorporated Heat detection for animals including cows
US5881673A (en) 1997-09-25 1999-03-16 Beach; Mark Heat detection system
WO2006077589A2 (fr) * 2005-01-19 2006-07-27 Scr Engineers Ltd. Procede et dispositif de detection de chaleurs
US7137359B1 (en) 2005-11-08 2006-11-21 Braden Joe T Estrus detector
US7230535B2 (en) 2003-07-17 2007-06-12 Jackson Iii William R Method and apparatus for monitoring breeding behavior
CA2648800A1 (fr) * 2006-03-03 2007-09-13 Fort Supply Ip, Llc Systeme de gestion de groupe social et methode connexe
US7927287B2 (en) 2006-11-09 2011-04-19 Mark Anderson Herd management technology
US7992521B2 (en) 2003-12-19 2011-08-09 Centre International d'Etudes Superieures en Sciences Agronomiques de Montepelliar (Montepellier Supagro) Method and device for automatically detecting mating of animals
US8066179B2 (en) 2006-11-10 2011-11-29 Breedcare Pty Ltd. Livestock breeding and management system
US8420398B2 (en) 2010-08-20 2013-04-16 Oratel Diagnostics, Llc Estrous cycle monitoring by color response
US8662021B2 (en) 2005-04-29 2014-03-04 Delaval Holding Ab Detecting method and arrangement for dairy cattle
US20140107434A1 (en) 2006-04-18 2014-04-17 Iti Scotland Limited Method and System for Monitoring the Condition of Livestock
US20140121558A1 (en) 2012-10-25 2014-05-01 Maria Victoria ALONSOPEREZ LANZA Device and method for the early detection of cattle physiological variables and location in a remote and autonomous way
US8949034B2 (en) 2011-03-17 2015-02-03 Technologies Holdings Corp. System and method for health monitoring using real-time location
US20150202033A1 (en) 2011-03-17 2015-07-23 Technologies Holdings Corp. System and method for estrus detection using real-time location
WO2015177741A1 (fr) * 2014-05-23 2015-11-26 Consejo Nacional De Investigaciones Científicas Y Técnicas (Conicet) Procédé de surveillance, de quantification et d'évaluation des activités de pâturage et de rumination réalisées par des ruminants et dispositif de mise en œuvre correspondant
US20160165851A1 (en) 2013-06-14 2016-06-16 Dairymaster A method, a device and a system for detecting a state of an animal

Patent Citations (24)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US708375A (en) 1902-03-15 1902-09-02 Theodore Wick Lowe Windmill.
US4247758A (en) 1979-11-15 1981-01-27 Rodrian James A Animal identification and estrus detection system
US4618861A (en) 1985-03-20 1986-10-21 Cornell Research Foundation, Inc. Passive activity monitor for livestock
US4846106A (en) 1985-06-06 1989-07-11 Cowtronics, Inc. Method and apparatus for detecting standing heat in cattle
US5111799A (en) 1990-03-28 1992-05-12 Washington State University Research Foundation, Inc. Estrous detection systems
US5542431A (en) 1993-06-30 1996-08-06 Ddx Incorporated Heat detection for animals including cows
US5881673A (en) 1997-09-25 1999-03-16 Beach; Mark Heat detection system
US7230535B2 (en) 2003-07-17 2007-06-12 Jackson Iii William R Method and apparatus for monitoring breeding behavior
US7992521B2 (en) 2003-12-19 2011-08-09 Centre International d'Etudes Superieures en Sciences Agronomiques de Montepelliar (Montepellier Supagro) Method and device for automatically detecting mating of animals
WO2006077589A2 (fr) * 2005-01-19 2006-07-27 Scr Engineers Ltd. Procede et dispositif de detection de chaleurs
US7878149B2 (en) 2005-01-19 2011-02-01 Scr Engineers Ltd. Method and device for detecting estrus
US8662021B2 (en) 2005-04-29 2014-03-04 Delaval Holding Ab Detecting method and arrangement for dairy cattle
US7137359B1 (en) 2005-11-08 2006-11-21 Braden Joe T Estrus detector
CA2648800A1 (fr) * 2006-03-03 2007-09-13 Fort Supply Ip, Llc Systeme de gestion de groupe social et methode connexe
US8979757B2 (en) 2006-04-18 2015-03-17 Iti Scotland Ltd. Method and system for monitoring the condition of livestock
US20140107434A1 (en) 2006-04-18 2014-04-17 Iti Scotland Limited Method and System for Monitoring the Condition of Livestock
US7927287B2 (en) 2006-11-09 2011-04-19 Mark Anderson Herd management technology
US8066179B2 (en) 2006-11-10 2011-11-29 Breedcare Pty Ltd. Livestock breeding and management system
US8420398B2 (en) 2010-08-20 2013-04-16 Oratel Diagnostics, Llc Estrous cycle monitoring by color response
US8949034B2 (en) 2011-03-17 2015-02-03 Technologies Holdings Corp. System and method for health monitoring using real-time location
US20150202033A1 (en) 2011-03-17 2015-07-23 Technologies Holdings Corp. System and method for estrus detection using real-time location
US20140121558A1 (en) 2012-10-25 2014-05-01 Maria Victoria ALONSOPEREZ LANZA Device and method for the early detection of cattle physiological variables and location in a remote and autonomous way
US20160165851A1 (en) 2013-06-14 2016-06-16 Dairymaster A method, a device and a system for detecting a state of an animal
WO2015177741A1 (fr) * 2014-05-23 2015-11-26 Consejo Nacional De Investigaciones Científicas Y Técnicas (Conicet) Procédé de surveillance, de quantification et d'évaluation des activités de pâturage et de rumination réalisées par des ruminants et dispositif de mise en œuvre correspondant

Non-Patent Citations (8)

* Cited by examiner, † Cited by third party
Title
ARCIDIACONO C; PORTO S; MANCINO M; G CASCONE: "A threshold-based algorithm for the development of inertial sensor-based systems to perform real-time cow step counting in free-stall barns", BIOSYSTEMS ENGINEERING, vol. 153, 2017, pages 99 - 109, XP029859945, DOI: doi:10.1016/j.biosystemseng.2016.11.003
CHELOTTI J; VANRELL S; MILONE D; UTSUMI S; GALLI J; RUFINER H; L GIOVANIN: "A real-time algorithm for acoustic monitoring of ingestive behavior of grazing cattle", COMPUTERS AND ELECTRONICS IN AGRICULTURE, vol. 127, 2016, pages 64 - 75, XP029709515, DOI: doi:10.1016/j.compag.2016.05.015
CLAPHAM W; FEDDERS J; BEEMAN K; J NEEL: "Acoustic monitoring system to quantify ingestive behavior of free-grazing cattle", COMPUTERS AND ELECTRONICS IN AGRICULTURE, vol. 76, 2011, pages 96 - 104, XP028196902, DOI: doi:10.1016/j.compag.2011.01.009
DIOSDADO J; BARKER Z; HODGES H; AMORY J; CROFT D; BELL N; E CODLING: "Classification of behaviour in housed dairy cows using an accelerometer-based activity monitoring system", ANIMAL BIOTELEMETRY, vol. 3, no. 1, 2015, pages 3 - 15
MILONE D; GALLI J; CANGIANO C; RUFINER H; E LACA: "Automatic recognition nof ingestive sounds of cattle based on hidden Markov models", COMPUTERS AND ELECTRONICS IN AGRICULTURE, vol. 87, 2011, pages 51 - 55, XP028424384, DOI: doi:10.1016/j.compag.2012.05.004
NAVON S; MIZRACH A; HETZRONI A; E UNGAR: "Automatic recognition of jaw movements in free-ranging cattle, goats and sheep, using acoustic monitoring", BIOSYSTEMS ENGINEERING, vol. 114, no. 4, 2013, pages 474 - 483, XP028989742, DOI: doi:10.1016/j.biosystemseng.2012.08.005
NIELSEN L; PEDERSEN A; HERSKIN M; L MUNKSGAARD: "Quantifying walking and standing behaviour of dairy cows using a moving average based on output from an accelerometer", APPLIED ANIMAL BEHAVIOUR SCIENCE, vol. 127, no. 1, 2010, pages 12 - 19
QASEM L; CARDEW A; WILSON A; GRIFFITHS I; HALSEY L; SHEPARD E; GLEISS A; R WILSON: "Tri-axial dynamic acceleration as a proxy for animal energy expenditure; should we be summing values or calculating the vector?", PLOS ONE, vol. 7, no. 2, 2012, pages e31187

Cited By (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11617352B2 (en) 2018-01-23 2023-04-04 William R. Jackson, III Method and apparatus for detection of estrus and optimal time for embryo transfer or artificial insemination in animals
CN109122371A (zh) * 2018-08-23 2019-01-04 贵州谷丰生态农业发展有限责任公司 一种肉牛放养用定位装置
US11574251B2 (en) 2019-09-12 2023-02-07 Performance Livestock Analytics, Inc. Livestock and feedlot data collection and processing using UHF-band interrogation of radio frequency identification tags for feedlot arrival and risk assessment
US12159208B2 (en) 2019-09-12 2024-12-03 Romance Livestock Analytics, Inc. Livestock and feedlot data collection and processing using UHF-band interrogation of radio frequency identification tags for feedlot arrival and risk assessment
WO2021050775A1 (fr) * 2019-09-12 2021-03-18 Performance Livestock Analytics, Inc. Collecte et traitement de données de bétail et de parc d'engraissement à l'aide d'une interrogation de bande uhf d'étiquettes d'identification radiofréquence d'arrivée en parc d'engraissement et d'évaluation de risque
US11055633B2 (en) 2019-09-12 2021-07-06 Performance Livestock Analytics, Inc. Livestock and feedlot data collection and processing using UHF-band interrogation of radio frequency identification tags for feedlot arrival and risk assessment
US11763205B2 (en) 2019-09-12 2023-09-19 Performance Livestock Anayltics, Inc. Livestock and feedlot data collection and processing using UHF-band interrogation of radio frequency identification tags for feedlot arrival and risk assessment
US20210295951A1 (en) * 2020-03-20 2021-09-23 Kpn Innovations, Llc Artificial intelligence methods and systems for generating zoological instruction sets from biological extractions
US11942207B2 (en) * 2020-03-20 2024-03-26 Kpn Innovations, Llc Artificial intelligence methods and systems for generating zoological instruction sets from biological extractions
WO2021234490A3 (fr) * 2020-05-19 2022-01-06 Agverse Technologies Private Limited Système et procédé de gestion de nutrition et de détection de chaleurs et d'insémination chez des animaux producteurs de lait
CN111870387A (zh) * 2020-08-04 2020-11-03 青岛农业大学 多功能母猪发情检测装置
RU2757683C1 (ru) * 2020-09-22 2021-10-20 Общество с ограниченной ответственностью "Научно-производственная компания "Биосенсорика" Способ определения состояния половой охоты сельскохозяйственного животного
WO2022115916A1 (fr) * 2020-12-04 2022-06-09 Finchain.Ai Pty Ltd Surveillance et gestion de bétail
WO2022154661A1 (fr) * 2021-01-18 2022-07-21 Nedap N.V. Procédé et système de détection d'œstrus chez un mammifère
NL2027332B1 (en) * 2021-01-18 2022-07-25 Nedap Nv Method and system for estrus detection of a mammal
US20240065233A1 (en) * 2021-01-18 2024-02-29 Nedap N.V. Method and system for estrus detection of a mammal

Also Published As

Publication number Publication date
AR112047A1 (es) 2019-09-18

Similar Documents

Publication Publication Date Title
WO2018109725A1 (fr) Procédé et dispositif de détection de chaleurs chez un ruminant
US10470441B2 (en) Animal monitor
Deniz et al. Embedded system for real-time monitoring of foraging behavior of grazing cattle using acoustic signals
US9615777B2 (en) System and method for monitoring of activity and fall
Jurdak et al. Camazotz: multimodal activity-based GPS sampling
CN105091938B (zh) 畜禽健康状况监测方法及系统
US10288740B2 (en) Position tracking method and apparatus
US20170325426A1 (en) A Method and Device for Remote Monitoring of Animals
CN105342623A (zh) 智能跌倒监护装置及其处理方法
CN102056478A (zh) 移动对象的检测
US12127531B2 (en) Livestock age verification system
CN113973739B (zh) 一种多功能的牛用电子耳标
JP5543317B2 (ja) 牛の発情検知方法
EP3836781B1 (fr) Procédé, appareil et système de détection d'événements de miction pour animaux d'élevage
EP4241560B1 (fr) Systèmes et procédés pour déterminer la consommation alimentaire de bétail
Sommer et al. From the lab into the wild: Design and deployment methods for multi-modal tracking platforms
Dulyala et al. Classification system for estrus behavior of cow using an accelerometer
US20240008456A1 (en) Wearable device for pet
le Roux et al. Energy benefits of on-board behaviour classification for animal-borne sensor applications
AURASOPON Dynamic Time Warping for classifying cattle behaviors and reducing acceleration data size
CN115802886A (zh) 管理装置、管理方法以及管理程序
CN113273511A (zh) 动物监测装置及方法
CN204166554U (zh) 基于rfid辅助标签定位的野生翠鸟捕食习性数据采集装置
KR20200105080A (ko) ＬＰＷＡＮ(ＬｏＲａ)와 ＢＬＥ를 활용한 소 신체 정보 수집 시스템
US20250213137A1 (en) Ring assembly for animal monitoring

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17832338

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17832338

Country of ref document: EP

Kind code of ref document: A1