CN-122004134-A - Self-adaptive intelligent control system of milking robot

CN122004134ACN 122004134 ACN122004134 ACN 122004134ACN-122004134-A

Abstract

The invention provides a self-adaptive intelligent control system of a milking robot, and relates to the technical field of robots. The system performs distributed cooperative control based on a robot operating system, and each module operates in the form of independent functional nodes and realizes data interaction and state synchronization through an ROS message mechanism. The self-adaptive intelligent adjustment strategy is realized through the real-time updating of the moving target point position of the manipulator and the real-time obstacle avoidance adjustment. By adopting a visual recognition and coordinate calibration two-way line fusion algorithm, the real-time tracking and high-precision positioning of a dynamic target (cow nipple) are realized. The control system of the invention adopts a closed-loop control logic of 'recognition-decision-execution-feedback', and constructs an integrated closed-loop flow of 'recognition-cleaning-stimulation-adsorption-milking-disinfection'.

Inventors

ZHANG GUOFEI
SHANG FEI
LI WEI
Yan wanji
LIU YUNYI
XU ZHIHAO
LI LEI
WANG HONG

Assignees

内蒙古欧牧机械设备有限公司

Dates

Publication Date: 20260512
Application Date: 20260312

Claims (10)

1. The self-adaptive intelligent control system of the milking robot is characterized by comprising an upper computer control unit, a dairy cow identification module, a visual identification module, a mechanical arm motion control module, a sensing monitoring module and a lower relay execution module, wherein each module operates in an independent node mode and comprises a dairy cow number reading node, a dairy cow nipple identification node, a mechanical arm control node, a flowmeter node, a serial port node and an interface interaction node, and real-time interaction and synchronization of data and control information are realized through an ROS multi-node communication mechanism; The milk cow identification module comprises an RFID read-write device and a milk cow data management unit, wherein the RFID read-write device is arranged at the entrance of a milking place, the electronic ear tag of the milk cow is automatically read when the milk cow approaches, the identified ID number is transmitted to the upper computer control unit through a serial port, and the milk cow data management unit is used for recording and managing the vital signs of the milk cow, including the milking times, the milking time, the milking quantity and the nipple state; the visual recognition module adopts a depth camera to complete cow nipple recognition and space positioning, the depth camera is arranged on a flange plate at the tail end of the mechanical arm, an optical axis is provided with an upper elevation angle of 10-15 degrees relative to the normal line of the flange plate, and RGB-D image data of a cow breast area are acquired; the mechanical arm motion control module consists of a servo-driven multi-degree-of-freedom mechanical arm, and a milk cup executing assembly and a depth camera are arranged at the tail end of the mechanical arm to realize point position updating and milk cup installation; the lower relay execution module controls the electromagnetic valve group, the vacuum pump and the pulsator through the relay board to realize the execution actions of the milk cup, including standing upside down, vacuum and pulsation control, standing upside down of the rolling brush frame, starting and stopping of the medicated bath spray valve, switching of pipelines, and different milk treatment and milk discarding; The sensing monitoring module comprises a TOF distance sensor and a flowmeter, wherein the TOF sensor is arranged around a milk tray at the tail end of the manipulator, and is used for realizing self-adaptive obstacle avoidance, flow monitoring and negative pressure safety protection; The upper computer control unit is used as a system central control node, operates the ROS system based on the NVIDIA development platform, is connected with the peripheral communication module through the Ethernet interface, and simultaneously establishes high-speed serial port or CAN bus communication with the milk cow identification module, the visual identification module, the mechanical arm motion control module, the sensing monitoring module and the lower relay execution module to realize multi-module synchronous control.
2. The self-adaptive intelligent control system of the milking robot according to claim 1 is characterized in that the point position updating is achieved by converting the nipple position of a cow into point position information under a manipulator base coordinate system, and specifically comprises the steps of (1) obtaining a rotation matrix R1 and a translation vector T1 between a camera coordinate system and a manipulator end coordinate system by means of a hand eye calibration technology, converting three-dimensional coordinates of the cow nipple under a camera view angle into the manipulator end coordinate system by means of the conversion matrix (R1, T1), and then achieving point position calculation of the manipulator end for moving a designated point on a milk disk to an actual nipple position by means of setting compensation values, so that a manipulator target position is obtained, and (2) converting the difference value, namely the actual distance, into the point position information under the manipulator coordinate system according to the difference value of the position information of the cow nipple under the camera view angle and the camera view angle when the final cow nipple is located at the designated point, obtaining a manipulator target position, and updating the point position when four nipple positions are recognized.
3. The self-adaptive intelligent control system of a milking robot according to claim 2, wherein the robot arm motion control module sets two distance thresholds, specifically including a scram threshold and a dangerous distance threshold, when the scram threshold is reached, the robot arm immediately stops the current motion state, i.e. powering down, and enters an adjustment state, moves in the opposite direction to a safe area according to the motion dangerous direction, returns to the starting point along the safe path from the current position if the dangerous distance threshold is triggered in the adjustment state, returns to the last step of work flow after moving to the safe distance if the dangerous distance threshold is not triggered, and cancels the scram state.
4. An adaptive intelligent control system for a milking robot as claimed in claim 3, further comprising a central database, wherein the central database is continuously updated with cow origin databases, including pastures for directly collecting cow self-status and feeding status dimension data.
5. The self-adaptive intelligent control system of the milking robot according to claim 4, wherein the milk cow identification system adopts an RFID technology to realize individual milk cow identification and data binding, specifically correlates milk cow identity with milk yield, health index, nipple position information, milking times and body weight of the milk cow to form a complete data file, meanwhile, the milk cow identification system judges whether the milk cow is first milking and meets milking conditions according to milk cow identification results to realize intelligent decision, namely, the execution of milking according to the body condition of the milk cow, and in addition, if the nipple position information is not needed for the first milking, the operations of medicated bath, rolling brush cleaning and milking are needed to be carried out after the identification of the identification position, and if the number of times of milking is two or more, weighting adjustment is carried out according to the data file, so that the consumption time of the identification process is reduced.
6. The self-adaptive intelligent control system of the milking robot of claim 5, wherein the milk cow identification system provides an open information interconnection architecture, confirms the unique identity of the milk cow through an electronic ear tag, a visual identification or an RFID module before the milk cow enters a fence, and establishes a digital file containing body type, nipple coordinates, health indexes, milk production records and feeding information in a central database; The milking manipulator motion control system is responsible for converting the identified nipple position information into a manipulator working space coordinate system, planning a manipulator motion track and realizing accurate positioning and motion control of the tail end of the manipulator to a target point; The lower relay control system receives an instruction from a serial port receiving and transmitting node of the control unit and controls the on-off state of the relay, so that the standing and inverting action of the milk cup, the start and stop control of the vacuum pump and the pulsator are realized, the distance between the adjacent sides of the mechanical arm can be monitored by the lower relay control system at high frequency, and the distance is fed back to the mechanical arm motion control module to realize the self-adaptive obstacle avoidance function; in the milking process, the manipulator is used as an executing mechanism to bear the realization of the movement function of the milk tray at the tail end, an inverse kinematics model of the manipulator is established according to the relevant size of the manipulator, the joint angles and the extending length of the connecting piece can be solved by giving a designated tail end position and posture, and then the accurate movement point function of the manipulator is realized through a manipulator controller, and the inverse kinematics model utilizes a D-H analysis method to solve the tail end position of the state of the manipulator into the rotation angle of each shaft joint according to the information of the joints and the arm length of the manipulator.
7. The self-adaptive intelligent control system of the milking robot according to claim 6, wherein the self-adaptive intelligent control system performs nipple positioning identification through a depth camera and performs real-time obstacle avoidance adjustment through a TOF sensor to realize stable identification and positioning of nipple points; The method comprises the steps of acquiring an image containing infrared and depth information through a depth camera, separating foreground noise and background noise through a fusion algorithm, adopting an image defogging algorithm based on contrast gradient to recover nipple edge details aiming at water vapor blurring, automatically triggering a lens dirt scraper to remove dirt when the image quality does not reach the standard, and maintaining visual definition; the method comprises the steps of identifying a cow nipple, extracting nipple features and spatial point positions by adopting a convolutional neural network, directly guiding a mechanical arm to align by an upper computer control unit if four nipples are identified, calling a geometric template if partial shielding exists to only identify three nipples, estimating a fourth nipple position by combining three-point coordinates and individual historical data based on the characteristic that cow nipples are commonly distributed in an isosceles trapezoid, and outputting after weight correction; If the identifiable point positions are less than three, the upper computer control unit starts an active exploration mode, the mechanical arm is controlled to conduct small-amplitude search within a safety range, the depth camera continuously collects data and updates the identification result, once a new nipple is found, the search is stopped immediately, the target layout is recalculated, complete information is recovered gradually, and vision-movement autonomous correction is achieved.
8. The self-adaptive intelligent control system of a milking robot according to claim 7, wherein a dynamic synchronous self-adaptive control mechanism of the mechanical arm and the nipple position is provided for solving the problem that the PID control cannot compensate the disturbance caused by the vacuum pulsation and the milk flow mutation in real time; the method comprises the steps of dynamically coupling the motion control and the vacuum adjustment of the mechanical arm by monitoring the tail end position and the nipple adsorption state of the mechanical arm in real time to form bivariate coordination control of the position and the flow so as to maintain adsorption stability and negative pressure balance under an unsteady state working condition; in the aspect of mechanical arm motion control, an upper computer control unit calculates the space error between the tail end of the mechanical arm and a nipple in real time, a self-adaptive gain adjustment mechanism is introduced into the traditional PID control for movement or shaking of the dairy cows in the milking process, PID parameters are dynamically adjusted according to displacement and vacuum change rate, so that the controller keeps optimal response under different disturbance, once the posture change of the dairy cows is detected, the upper computer control unit drives the mechanical arm to conduct micro compensation motion and synchronously follows the dynamic position of the nipple, and the stable contact between a breast cup and the nipple is ensured.
9. The self-adaptive intelligent control system of the milking robot according to claim 8, wherein the upper computer control unit introduces a fuzzy logic regulator, and the fuzzy logic regulator automatically judges the stability of the current working condition according to the real-time monitored nipple displacement, vacuum fluctuation and milk flow rate change and switches control gain and response rate parameters according to the stability; When abnormal fluctuation of any vacuum milk sucking channel is detected, the upper computer control unit immediately triggers the protection logic, reduces the adsorption force and pauses the movement of the mechanical arm, and after the emulsion flow read by the flowmeter is restored to be stable, the upper computer control unit drives the mechanical arm to automatically resume operation, so that the whole process safety self-resuming operation is realized.
10. An adaptive intelligent control method of a milking robot, implemented by an adaptive intelligent control system of a milking robot as claimed in claim 1, comprising the steps of: S1, initializing and waiting a system, wherein after an upper computer control unit is started, a mechanical arm motion control module is automatically electrified and controls the mechanical arm to move to a preset safe return-to-zero position; s2, after receiving a starting instruction, the system firstly acquires identity information of the dairy cow through a code reader or an RFID technology, and judges whether the dairy cow is a new dairy cow for primary milking or a old dairy cow for secondary operation according to the identification result; for new cows, starting a complete initialization process comprising visual calibration and milk cup preparation, namely a jumping step S3, and for old cows, directly calling stored historical data to enter a core operation link, namely a jumping step S5; S3, for a new cow, the mechanical arm moves to a fixed photographing position preset according to a working environment, and image capturing and archiving are carried out on a breast area of the cow by the depth camera, then the mechanical arm moves to a position which is positioned right below the abdomen of the cow and enables the milk pan to be parallel to the cow, the depth camera continuously collects images at the position, coordinate solving and coordinate self-adaptive correction are achieved through fusion of a space mapping method based on a calibration matrix and a dynamic compensation method based on a visual difference value, and coordinate sets of four teats are obtained through real-time calculation and optimization; S4, sequentially converting the four teat cups from an upright standby state to an inclined posture which is biased towards the camera and does not shade the camera by controlling the electromagnetic valve group; s5, performing medicated bath operation, namely sequentially positioning the mechanical arm below four target teats according to the calculated coordinates, sending an instruction by the upper computer control unit after reaching each designated position, starting a medicated bath spray valve, spraying disinfectant for a period of seconds on the teats, and automatically closing after finishing; s6, after the medicated bath is completed, rolling and brushing cleaning operation is carried out; the upper computer control unit controls the rolling brush support to rise and start to rotate on each point, and simultaneously the mechanical arm performs up-and-down micro motion in the vertical direction to simulate manual wiping action to thoroughly and physically clean the nipple; S7, entering a cup and milking stage, wherein the mechanical arm carries the milk cup to sequentially move to calculated cup coordinate points, namely, the calculated cup points under the mechanical arm coordinate system are relative to the positions of the teats under the mechanical arm coordinate system, and the milk cup can be adsorbed and attached to the teats by executing rope loosening action; S8, in the milking process, the upper computer control unit continuously monitors real-time data returned by the four flow meters, and independently judges and decides the state of each milk area, when the flow is maintained within a preset range, the current parameters are maintained to continuously work, when the flow rate of a certain milk area is detected to be lower than the preset range, the working frequency of a pulsator is increased, once the flow rate of the certain milk area is judged to disappear, the cup removing operation is immediately carried out for the milk area, and the cup removing operation comprises the steps of closing milk valves, stopping vacuum and recovering milk cups until all the milk areas complete milking; s9, after milking is completed completely, once dipping operation is executed again, postpartum disinfection is carried out on the teats, finally, the mechanical arm carries the teatcup sets to return to the zero position safely, the self-adaptive intelligent control system resets all internal state marks, and time consumption and yield of the operation are uploaded to the dairy cow data management unit for production analysis.

Description

Self-adaptive intelligent control system of milking robot Technical Field The invention relates to the technical field of robots, in particular to a self-adaptive intelligent control system of a milking robot. Background With the continuous increase of the consumption of the global dairy products, the dairy cow breeding industry is facing three pain points of 'labor waste, high cost and fluctuation of quality'. Traditional fishbone type or rotary disc type milking parlor relies on artificial sleeve cups and inspection, a skilled worker can fall in the morning for two shifts, labor intensity is high, mastitis is easily caused by operation difference, and the number of somatic cells of raw milk exceeds standard. The EU 2025 regulation has reduced the limit of somatic cell number from 40 ten thousand to 30 ten thousand/mL, and the pour pasture improves the sanitation grade by an accurate and unmanned means. Meanwhile, the large-scale acceleration of dairy cow stock, the proportion of ten thousands of pastures is more than 35%, the efficient milking of thousands of dairy cows is required to be completed every shift, and the manual mode can not meet the 24-hour continuous production requirement. Although milking robots introduced by European and American manufacturers are already on the land in domestic pastures, there are still "ocean proposal water and soil not taking": (1) The static model is difficult to adapt to the working conditions of large difference of Chinese Holstein cow body types and unstable gait, and the rate of missing extrusion and cup falling is more than 8%; (2) The single vision positioning fails to identify under the interference of fecal pollution and water vapor, so that the mechanical arm collides with the nipple, the stress of the dairy cows is caused, and the milk yield is reduced by 10% -15%; (3) The existing PID control cannot compensate disturbance caused by vacuum pulsation and milk flow mutation on line, and the vacuum fluctuation of the nipple end is +/-6 kPa, which is higher than national standard +/-3 kPa, so that the risk of mammitis is increased; (4) The system is closed, a domestic TMR, oestrus monitoring and DHI big data platform cannot be connected, an information island is formed, and intelligent control of the whole life cycle is difficult to realize. Therefore, developing a set of self-adaptive intelligent control system with self-learning and self-optimizing capabilities for the milking robot has become the key to break through bottleneck, realize domestic replacement of high-end agricultural machinery and ensure the milk industry to be happy. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a self-adaptive intelligent control system of a milking robot, which aims to construct a multi-level control system with perception intelligence, behavior intelligence and decision intelligence so as to realize high-precision milking operation and full life cycle data management in a complex dynamic environment. The invention relates to an agricultural service type electromechanical integrated device, in particular to a self-adaptive intelligent milking robot, which realizes the automatic, accurate and efficient milking operation of cows. The device integrates multi-disciplinary technologies such as machinery, electricity, computer vision, information processing and the like, and can complete the whole process of automatic milking from cow identification, positioning and nipple detection of the mechanical arm under the condition of no human intervention. The technical scheme of the invention is as follows: On the one hand, the invention provides a self-adaptive intelligent control system of a milking robot, which comprises an upper computer control unit, a dairy cow identification module, a visual identification module, a mechanical arm motion control module, a sensing monitoring module and a lower relay execution module, wherein each module operates in an independent node mode and comprises a dairy cow number reading node, a dairy cow nipple identification node, a mechanical arm control node, a flowmeter node, a serial port node and an interface interaction node, and real-time interaction and synchronization of data and control information are realized through an ROS multi-node communication mechanism. The milk cow identification module comprises an RFID read-write device and a milk cow data management unit, wherein the RFID read-write device is arranged at the entrance of a milking place, the electronic ear tag of the milk cow is automatically read when the milk cow approaches, the identified ID number is transmitted to the upper computer control unit through a serial port, and the milk cow data management unit is used for recording and managing the vital signs of the milk cow, including the milking times, the milking time, the milking quantity and the nipple state; the visual recognition module adopts a depth camera to complete cow nipple re