Human-machine collaborative assembly robot control system and control method

Abstract

The invention discloses a control system and a control method for a human-machine collaborative assembly robot. The control system comprises a network main control layer, which is communicatively connected with a master-slave drive control layer, which receives control instructions from a PC host computer and collects and feeds back data information from an execution component layer, which is communicatively connected with the execution component layer, which completes overall motion control of the assembly robot and feeds back data information from the execution component layer. The control method comprises the following steps: a human-machine interaction software interface module and a control cabinet operation panel module receive control instruction information; a force / torque sensor module receives collaborative control instructions; target motion information is calculated; an encoder code disk value corresponding to a servo motor is calculated; a control signal is sent to the motor; and the posture of a terminal moving platform is accurately adjusted until the assembly operation is completed. The invention has low development difficulty and low cost, meets portability requirements, has high real-time performance and good reliability, and realizes adaptive control of human-machine collaborative assembly.

Description

Technical Field

[0001] The invention belongs to the technical field of assembly robot control, and in particular relates to a human-machine collaborative assembly robot control system and a control method. Background Art

[0002] The use of semi-automatic assembly technology and equipment for human-machine collaboration is of great significance for ensuring assembly quality and improving assembly efficiency. The fundamental reason for the emergence of collaborative robots is closely related to the inability of traditional industrial robots to meet the complex production and manufacturing requirements. Through human-machine collaboration, the quality of work can be further guaranteed, while improving the comfort of workers' work, and ultimately achieving safe, flexible, and efficient assembly work for human-machine collaboration, solving the application needs of low-cost, high-efficiency, flexible, and complex work automation that traditional industrial robots are difficult to cope with.

[0003] The software-based open structure of the current robot control system type is based on PC and runs software platform servo control under the computer operating system, realizing the software-based control solution. It has strong openness, can meet the diverse needs of users, and has a very broad development and application prospect.

[0004] However, due to the high requirements of the human-machine collaborative assembly field for the portability, reliability and advanced control technology of the robot system, the control system and control method of the existing collaborative assembly robot still have the following shortcomings:

[0005] First, the development cost is high and the portability requirement is difficult to meet. For example, the robot control system described in patent CN109202722A uses a purchased motion control card as the main controller to implement the corresponding functions. The control system cost is high, and this type of system is large in size and inconvenient.

[0006] Second, it is difficult to ensure the real-time performance of the system, and the reliability is poor. For example, the robot control system based on the serial communication protocol described in patent CN110757460A, this communication method is usually carried out in the form of master station polling, and the delay characteristics of the system will seriously affect the real-time performance of the system, and the reliability is poor.

[0007] Third, there is a lack of integrated control technology and methods for human-machine collaborative assembly. For example, the admittance control algorithm described in patent CN109249394B only realizes the collaborative control of robots based on guiding force and speed, and selects the admittance coefficient by simply dividing the interval, which lacks adaptability and flexibility. Summary of the invention

[0008] The present invention is proposed to solve the problems existing in the prior art, and its purpose is to provide a human-machine collaborative assembly robot control system and a control method.

[0009] The technical solution of the present invention is: a human-machine collaborative assembly robot control system, including a network main control layer, the network main control layer is communicatively connected with a master-slave drive control layer, the master-slave drive control layer receives and executes control instructions of a PC host computer in the network main control layer and collects and feeds back data information of an execution component layer, the master-slave drive control layer is communicatively connected with an execution component layer, the execution component layer completes the overall motion control of the assembly robot, and the execution component layer feeds back data information at the same time.

[0010] Furthermore, the network main control layer includes a human-computer interaction software interface module for realizing human-computer interface interaction, the human-computer interaction software interface module is connected to the PLC main control system module, the PLC main control system module realizes the communication connection configuration and data interaction of the system hardware, and performs real-time control of the hardware, and the PLC main control system module is connected to the collaborative motion control algorithm module.

[0011] Furthermore, the master-slave drive control layer includes a communication coupling module, which is connected to the input-output module, which connects external input-output variables to the EtherCAT field bus network, and the input-output module is connected to the force / torque sensor module, the driver module, and the control cabinet operation panel module. The driver module realizes high-precision positioning movement of the servo motor, receives and parses relevant instructions of the external control motor, and sends control signals to the corresponding motor.

[0012] Furthermore, the PC host computer in the network master control layer is an EtherCAT master station, and the master-slave drive control layer is an EtherCAT slave station.

[0013] Furthermore, the driver module is used for high-precision positioning movement of the servo motor, the input port of the driver module receives and parses relevant instructions of the external control motor, and the output of the driver module sends a control signal to the corresponding driver.

[0014] Furthermore, the force / torque sensor module is used to obtain the force / torque information applied by the operator during human-machine collaboration, and the output end of the force / torque sensor module transmits the force / torque information to the input-output module.

[0015] Furthermore, the execution component layer includes a motor, which receives control information from a driver and executes, and the motor is connected to a servo motor module.

[0016] A control method for a human-machine collaborative assembly robot control system comprises the following steps:

[0017] ⅰ. The human-computer interaction software interface module and the control cabinet operation panel module receive the operator's control command information;

[0018] ⅱ. The force / torque sensor module receives the operator's collaborative control instructions;

[0019] ⅲ. Receive control command information and collaborative control command, and calculate the target motion information of the robot terminal moving platform according to the corresponding algorithm;

[0020] ⅵ. Obtain the rod length of each branch chain of the robot through the inverse solution algorithm of the robot mechanism kinematics, and convert it to the encoder disk value of the corresponding servo motor according to the lead pitch relationship of the ball screw;

[0021] ⅴ. The driver module receives the target value of the encoder and sends a control signal to the corresponding motor according to the currently selected control mode;

[0022] ⅵ. The robot's end moving platform is directly dragged to the vicinity of the target position by the operator, and the end moving platform's posture is further accurately adjusted through the human-machine interaction software interface module and the control cabinet operation panel module according to the actual working conditions until the assembly operation is completed;

[0023] ⅶ. During collaborative assembly, the operator may press the emergency stop protection button contained in the human-machine interaction software interface module and the control cabinet operation panel module at any time depending on the specific working conditions, or stop contact with the force / torque sensor module to stop the robot from moving.

[0024] Furthermore, the PLC main control system module is an ADS server, the collaborative motion control algorithm module is an ADS client, the ADS client sends an ADS request to the ADS server, and during the communication process, the client program stops executing until a response is returned by the ADS server.

[0025] Furthermore, the collaborative motion control algorithm module includes a joint space trajectory planning control algorithm and a human-machine collaborative adaptive admittance control algorithm;

[0026] The joint space trajectory planning control algorithm is implemented based on a trajectory planning control algorithm of cubic polynomial interpolation;

[0027] The human-machine collaborative adaptive admittance control algorithm is implemented based on a variable admittance control algorithm of machine reinforcement learning.

[0028] The beneficial effects of the present invention are as follows:

[0029] The invention has low development difficulty and low cost, and meets the portability requirement. The open structure of the invention is clear and reasonable, easy to operate, safe to run, saves space for motion control cards, reduces costs, gives the host computer real-time computing capabilities, adopts modular thinking to enhance the overall packaging of the system, reduces development difficulty, transmits data through network cables, and the system is portable.

[0030] The present invention has high real-time performance and good reliability. Compared with the traditional field bus, the present invention has high real-time performance and good time certainty, the transmitted information is mostly short frame information, the information exchange is frequent, and the fault tolerance is strong and the reliability is good.

[0031] The present invention can realize adaptive control of human-machine collaborative assembly. When the adaptive admittance control algorithm of the present invention is used for human-machine collaborative assembly operations, the system adjusts the admittance parameters in real time according to the interactive force information collected by the force / torque sensor installed at the end and the robot end motion information to enhance the flexibility and comfort of the collaboration, thereby improving the efficiency of collaborative assembly and reducing the labor intensity of workers. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a system structure block diagram of the control system in the present invention;

[0033] Figure 2 It is a structural block diagram of the human-computer interaction software interface module in the present invention;

[0034] Figure 3 It is a diagram of the robot reinforcement learning algorithm architecture in the present invention;

[0035] Figure 4 It is a flow chart of the variable admittance control algorithm based on machine reinforcement learning in the present invention;

[0036] in:

[0037] 1 Network master control layer 2 Master-slave drive control layer

[0038] 3 Execution component layer

[0039] 11 PLC main control system module 12 Human-computer interaction software interface module

[0040] 13 Collaborative motion control algorithm module

[0041] 21 Communication coupling module 22 Input and output module

[0042] 23 Force / torque sensor module 24 Driver module

[0043] 25 Control cabinet operation panel module 26 driver

[0044] 27 Electrical Components Module

[0045] 31 Motor 32 Servo Motor Module

[0046] 33 Assembly Robot

[0047] 121 Data exchange communication function 122 Robot motion control function

[0048] 123 Information collection and feedback function

[0049] 1211 Servo drive communication interface

[0050] 1212 Force / torque sensor communication interface

[0051] 1231 Servo motor action information collection

[0052] 1232 Force / torque sensor information acquisition and processing

[0053] 1233 Feedback of motion information of the robot's terminal moving platform. DETAILED DESCRIPTION

[0054] Hereinafter, the present invention will be described in detail with reference to the accompanying drawings and embodiments:

[0055] like Figures 1 to 4 As shown, a human-machine collaborative assembly robot control system includes a network main control layer 1, which is communicatively connected to a master-slave drive control layer 2. The master-slave drive control layer 2 receives and executes control instructions from a PC host computer in the network main control layer 1 and collects and feeds back data information of an execution component layer. The master-slave drive control layer 2 is communicatively connected to an execution component layer 3, which completes the overall motion control of the assembly robot and feeds back data information to the execution component layer 3.

[0056] The network main control layer 1 includes a human-computer interaction software interface module 12 for realizing human-computer interface interaction. The human-computer interaction software interface module 12 is connected to a PLC main control system module 11. The PLC main control system module 11 realizes the communication connection configuration and data interaction of the system hardware, and performs real-time control of the hardware. The PLC main control system module 11 is connected to a collaborative motion control algorithm module 13.

[0057] The master-slave drive control layer 2 includes a communication coupling module 21, which is connected to an input-output module 22. The input-output module 22 connects external input-output variables to the EtherCAT field bus network. The input-output module 22 is connected to a force / torque sensor module 23, a driver module 24, and a control cabinet operation panel module 25. The driver module 24 realizes high-precision positioning movement of the servo motor, receives and parses relevant instructions of the external control motor, and sends control signals to the corresponding motor.

[0058] The PC host computer in the network master control layer 1 is an EtherCAT master station, and the master-slave control layer 2 is an EtherCAT slave station.

[0059] The driver module 24 is used for high-precision positioning movement of the servo motor. The input port of the driver module 24 receives and analyzes relevant instructions of the external control motor, and the output of the driver module 24 sends a control signal to the corresponding driver 26.

[0060] The force / torque sensor module 23 is used to obtain the force / torque information applied by the operator during human-machine collaboration, and the output end of the force / torque sensor module 23 transmits the force / torque information to the input / output module 22 .

[0061] The execution component layer 3 includes a motor 31 , which receives control information from the driver 26 and executes. The motor 31 is connected to a servo motor module 32 .

[0062] Specifically, the network main control layer 1 is designed and developed based on the control software platform of the host computer, and is specifically composed of four parts: a PC host computer, a PLC main control system module 11, a collaborative motion control algorithm module 13 and a human-computer interaction software interface module 12.

[0063] The PC host computer is an EtherCAT master station, which is installed with Beckhoff TwinCAT3 automation industrial control software platform.

[0064] The PLC main control system module 11 has a built-in forward and inverse solution algorithm for the kinematics of the assembly robot.

[0065] The collaborative motion control algorithm module 13 is developed and designed based on the Visual Studio software environment, and is used for the system to solve the motion information of the robot terminal in real time according to feedback data, and has an embedded motion trajectory planning control algorithm and a human-machine collaborative adaptive admittance control algorithm.

[0066] The human-machine interaction software interface module 12 is implemented based on the HMI module in the Beckhoff TwinCAT3 platform and is used for human-machine information interaction.

[0067] At the same time, the PLC main control system module 11, the human-computer interaction software interface module 12 and the collaborative motion control algorithm module 13 are all located in the PC host station, which belongs to local communication. The ADS (Automation Device Specification) communication protocol based on TCP / IP is used for data exchange between the TwinCAT3 platform and the Visual Studio software environment.

[0068] Specifically, the master-slave drive control layer 2 is used to receive and execute control instructions from the PC host computer and collect data information from the feedback execution component layer. It is specifically composed of six parts: a communication coupling module 21, an input and output module 22, a driver module 24, a force / torque sensor module 23, a control cabinet operation panel module 25 and an electrical component module 27.

[0069] The input-output module 22 connects external input-output variables to the EtherCAT fieldbus network. The input-output module 22 specifically includes an analog quantity type module and a digital quantity type module.

[0070] The driver module 24 is used for high-precision positioning movement of the servo motor, receives and analyzes relevant instructions of the external control motor, and sends control signals to the corresponding driver 26.

[0071] More specifically, the driver module 24 includes six sets of servo motor drivers.

[0072] The force / torque sensor module 23 is used to obtain the force / torque information applied by the operator during human-machine collaboration, and specifically includes a set of six-dimensional force / torque sensors.

[0073] The control cabinet operation panel module 25 is used to receive the operator's control instructions, and specifically includes control hardware such as a self-locking / self-resetting button, a second-gear / third-gear knob, etc.

[0074] The electrical component module 27 is used to realize the logic control circuit of the perfect control hardware system, including control hardware such as contactors, DC power supplies, DC circuit breakers, intermediate relays, wiring terminals and cables.

[0075] Specifically, the execution component layer 3 is used to complete the motion control of the robot as a whole and feedback data information, and is specifically composed of an assembly robot 33 and a servo motor module 32:

[0076] The assembly robot 33 is a six-degree-of-freedom parallel robot, specifically comprising a moving platform, a static platform, and six motion branches, each of which is composed of two ball-joint kinematic pairs and a ball screw linear kinematic pair.

[0077] The servo motor module 32 specifically includes six sets of servo motors and corresponding motor brakes and absolute encoders.

[0078] The human-machine interaction software interface module 12 includes a data interaction communication function 121 , a robot motion control function 122 , and an information collection and feedback function 123 .

[0079] The data interactive communication function 121 includes a servo driver communication interface 1211 and a force / torque sensor communication interface 1212. The servo driver communication interface 1211 and the force / torque sensor communication interface 1212 realize the connection of corresponding interfaces.

[0080] The information collection and feedback function 123 includes servo motor motion information collection 1231, force / torque sensor information collection and processing 1232, and robot terminal moving platform motion information feedback 1233. Servo motor motion information collection 1231, force / torque sensor information collection and processing 1232, and robot terminal moving platform motion information feedback 1233 realize the collection of corresponding data.

[0081] A control method for a human-machine collaborative assembly robot control system comprises the following steps:

[0082] ⅰ. The human-computer interaction software interface module and the control cabinet operation panel module receive the operator's control command information;

[0083] ⅱ. The force / torque sensor module receives the operator's collaborative control instructions;

[0084] ⅲ. Receive control command information and collaborative control command, and calculate the target motion information of the robot terminal moving platform according to the corresponding algorithm;

[0085] ⅵ. Obtain the rod length of each branch chain of the robot through the inverse solution algorithm of the robot mechanism kinematics, and convert it to the encoder disk value of the corresponding servo motor according to the lead pitch relationship of the ball screw;

[0086] ⅴ. The driver module receives the target value of the encoder and sends a control signal to the corresponding motor according to the currently selected control mode;

[0087] ⅵ. The robot's end moving platform is directly dragged to the vicinity of the target position by the operator, and the end moving platform's posture is further accurately adjusted through the human-machine interaction software interface module and the control cabinet operation panel module according to the actual working conditions until the assembly operation is completed;

[0088] ⅶ. During collaborative assembly, the operator may press the emergency stop protection button contained in the human-machine interaction software interface module and the control cabinet operation panel module at any time depending on the specific working conditions, or stop contact with the force / torque sensor module to stop the robot from moving.

[0089] The PLC main control system module is an ADS server, and the collaborative motion control algorithm module is an ADS client. The ADS client sends an ADS request to the ADS server. During the communication process, the client program stops executing until a response is returned by the ADS server.

[0090] To facilitate the use of ADS communication in the computer high-level language Visual C++ and its environment, Beckhoff TwinCAT3 provides the TcAdsDll.dll component. The files required for linking and program development in Visual C++ include the dynamic link library TcAdsDll.dll, the function library TcAdsDll.lib, the header file TcAdsApi.h declaring ADS, and the header file TcAdsDef.h declaring structures and constants.

[0091] Specifically, the control system adopts the ADS communication synchronous reading and writing mode, which can return the data interaction results immediately. The PLC main control system module 11 acts as an ADS server, and the collaborative motion control algorithm module 13 acts as an ADS client. The ADS client sends an ADS request to the ADS server. During the communication process, the client program stops executing until a response is returned by the ADS server.

[0092] At the same time, it is hereby explained that the ADS client and the ADS server are both designed in the same PC host computer and belong to local communication.

[0093] The PLC main control system module 11 and the cooperative motion control algorithm module 13 perform data information exchange, and the specific process is as follows:

[0094] a. Define port variables, port address variables, and AMS address variables in the Visual Studio environment.

[0095] b. Open the ADS communication port by calling the AdsPortOpen() instruction, and automatically obtain the local address by calling the AdsGetLocalAddress() instruction, thereby establishing communication between the PLC main control system module 11 and the collaborative motion control algorithm module 13.

[0096] c. Set the port address pointer to port 851 of TwinCAT3.

[0097] d. Create a handle variable in the Visual Studio environment, define the control system variable to be read / written and the string variable corresponding to the control system variable.

[0098] e. By calling the AdsSyncReadWriteReq() instruction, data can be synchronously written to the ADS server and received from the ADS device. According to the ADS server address and the control system variable to be read / written, the handle size corresponding to the variable is automatically generated.

[0099] f. Data can be read synchronously from the ADS server by calling the AdsSyncReadReq() instruction.

[0100] g. Data can be written synchronously to the ADS server by calling the AdsSyncWriteReq() instruction.

[0101] Specifically, the collaborative motion control algorithm module 13 specifically includes a joint space trajectory planning control algorithm and a human-machine collaborative adaptive admittance control algorithm.

[0102] The joint space trajectory planning control algorithm is implemented based on a trajectory planning control algorithm of cubic polynomial interpolation, and the human-machine collaborative adaptive admittance control algorithm is implemented based on a variable admittance control algorithm of machine reinforcement learning.

[0103] Specifically, the joint space trajectory planning control algorithm includes the following steps:

[0104] a. The control system receives the operator's control instructions according to the selected current working mode, and obtains the position and posture information of the key points of the center point of the end moving platform of the assembly robot expected to reach the target point;

[0105] b. Through the mechanism kinematics inverse solution algorithm embedded in the PLC main control system module 11, the key point posture information of the end operation space is solved in sequence into the joint space path point information corresponding to the six motors to ensure smooth and impact-free operation of the motors.

[0106] c. Each motor performs trajectory interpolation based on the key path point information of the corresponding joint space and the cubic polynomial interpolation trajectory planning algorithm to obtain the motion position, velocity, and acceleration curve of the motor during this movement.

[0107] d. The servo motors all work in the position control mode, and the positions that each motor needs to move to at each cycle are sent in real time according to the clock cycle of the PLC main control system module 11 until the end moving platform moves to the desired target point.

[0108] Specifically, the human-machine collaborative adaptive admittance control algorithm includes the following steps:

[0109] a. Use the velocity information, velocity change rate information of the center point of the robot's terminal moving platform, the human-machine interaction force information and interaction force change rate information of the terminal six-dimensional force / torque sensor as observable state inputs to the reinforcement learning controller;

[0110] b. According to the Markov decision process of variable admittance parameter adjustment, the output admittance parameter damping value is adjusted in real time based on the selected machine reinforcement learning algorithm, and the parameter value is transmitted to the system admittance controller;

[0111] c. The cooperative displacement, velocity and acceleration information of the center point of the terminal moving platform generated by the human-machine interaction force are calculated based on the admittance controller;

[0112] d. The collaborative displacement, velocity and acceleration information are used as inputs to the robot control system to control the robot movement and complete the collaborative assembly task.

[0113] Yet another embodiment

[0114] As attached Figure 1 As shown, a control system of a human-machine collaborative assembly robot includes three aspects: a network main control layer 1, a master-slave drive control layer 2, and an execution component layer 3.

[0115] The network master control layer 1 is designed and developed based on the host computer industrial control software platform, and specifically consists of four parts: PC host computer master station, PLC master control system module 11, collaborative motion control algorithm module 13 and human-computer interaction software interface module 12:

[0116] The PC host station is installed with Beckhoff TwinCAT3 automation industrial control software platform, the PC kernel is Windows 10 operating system and must be equipped with a network card supported by TwinCAT3.

[0117] The PLC main control system module 11 is implemented based on the TwinCAT3 industrial control platform, and is used to realize the communication connection configuration and data interaction of the system hardware resources, while realizing real-time control of the hardware resources. It uses distributed clock technology to realize motion control and IO control functions. It is the main control core of the robot system, and it has an embedded assembly robot's mechanism kinematics forward and inverse solution algorithm.

[0118] The collaborative motion control algorithm module 13 is implemented based on the Visual Studio software environment. It is used for the system to solve the robot's subsequent motion information in real time according to feedback data. It has an embedded motion trajectory planning control algorithm and a human-machine collaborative adaptive admittance control algorithm. This module automatically runs / stops when the overall control system is started / shutdown.

[0119] The human-machine interaction software interface module 12 is implemented based on the HMI module in the Beckhoff TwinCAT3 platform and is used for human-machine information interaction. The operator can power on the robot system by clicking and touching the software interface, and select different working modes of the robot according to the logic algorithm of motion trajectory control to control the robot to better perform assembly operations. At the same time, the robot motion information can be collected in real time and feedback can be recorded, which is convenient for the operator to evaluate and judge the progress of the assembly process.

[0120] At the same time, the PLC main control system module 11, collaborative motion control algorithm module 13 and human-computer interaction software interface module 12 are all located in the PC host station, which belongs to local communication. The ADS (Automation Device Specification) communication protocol based on TCP / IP is used for data exchange between the TwinCAT3 platform and the Visual Studio software environment.

[0121] The master-slave drive control layer 2 is used to receive and execute control instructions from the PC host computer and collect data information from the feedback execution component layer. It is specifically composed of a communication coupling module 21, an input and output module 22, a driver module 24, a force / torque sensor module 23, a control cabinet operation panel module 25 and an electrical component module 27.

[0122] The communication coupling module 21 is used to couple multiple input and output modules together and connect them to the EtherCAT fieldbus network. The communication coupling module 21, the input and output module 22 and other connected control system hardware can be regarded as an EtherCAT slave station, which simplifies electrical wiring and saves control cabinet space.

[0123] The input and output module 22 is used to connect external input and output variables to the EtherCAT field bus network, so as to process input data and issue output instructions in the PLC main control system module 11, and specifically includes an analog type module and a digital type module.

[0124] The driver module 24 is used for high-precision positioning movement of the servo motor, receives and analyzes relevant instructions of the external control motor, and sends motor control signals to the corresponding motor. It uses an absolute encoder as a speed sensor and can control the servo motor through three control modes: position, speed and torque. Specifically, it includes six sets of servo motor drivers.

[0125] At the same time, based on the EtherCAT bus communication protocol, the PC host master station of the network master control layer is connected with the communication coupling module 21, input and output module 22, and driver module 24 of the master-slave drive control layer in sequence through a network cable, adopting a bus topology structure.

[0126] The force / torque sensor module 23 is used to obtain the force / torque information applied by the operator during human-machine collaboration. It can convert the analog signal into a digital signal through the embedded data acquisition system, so as to make corresponding processing in the network main control layer to realize human-machine collaborative control, and specifically includes a set of six-dimensional force / torque sensors.

[0127] At the same time, the input-output module 22 is connected to the force / torque sensor module 23 to collect input data of the force / torque signal applied by the operator.

[0128] The control cabinet operation panel module 25 is used to receive the operator's control instructions, and can achieve the same functions as the human-computer interaction software interface module 12. It can be used synchronously with the software interface or alone to complete the motion control of the robot, and specifically includes control hardware such as self-locking / self-resetting buttons, second-gear / third-gear knobs, etc.

[0129] At the same time, the input-output module 22 is connected to the equipment included in the control cabinet operation panel module 25 to perform input collection and output control.

[0130] The electrical component module 27 is used to realize the logic control circuit of the perfect control hardware system, including control hardware such as contactors, DC power supplies, DC circuit breakers, intermediate relays, wiring terminals and cables.

[0131] The execution component layer 3 is used for the robot to complete the motion operation as a whole and feedback information to the other two layers. It is specifically composed of an assembly robot 33 and a servo motor module 32.

[0132] The assembly robot 33 is a six-degree-of-freedom parallel robot, specifically comprising a moving platform, a static platform, and six motion branches, each of which is composed of two ball-joint kinematic pairs and a ball screw linear kinematic pair.

[0133] The servo motor module 32 specifically includes six sets of servo motors and corresponding motor brakes and absolute encoders.

[0134] At the same time, the servo motor module 32 is installed at the bottom end of the linear motion pair of the six motion branches of the assembly robot, and is connected to the ball joint motion pair fixed to the static platform through the motor external frame, and the other end of the linear motion pair is connected to the ball joint motion pair fixed to the dynamic platform.

[0135] As attached Figure 2 As shown, the functional structure of the human-machine interaction software interface module 12 includes a data interaction communication function 121 , an information collection and feedback function 123 , and a robot motion control function 122 .

[0136] The data interactive communication function 121 includes: a servo driver communication interface 1211 and a force / torque sensor communication interface 1212 .

[0137] The information collection and feedback function 123 includes: servo motor motion information collection 1231, force / torque sensor information collection and processing 1232, and robot terminal moving platform motion information feedback 1233.

[0138] The robot motion control function 122 includes: robot initial position return to zero, system enable / disable, robot spatial axis single degree of freedom step control, robot spatial axis single degree of freedom jog control, system emergency stop, and spatial trajectory planning control.

[0139] like Figure 3 As shown, the variable admittance control method based on machine reinforcement learning of the present invention is specifically based on the Deep Q-Learning Network (hereinafter referred to as DQN) algorithm, and includes the following steps:

[0140] a. Obtain the interaction force F collected by the force / torque sensor module 23 h Information, through the host computer processing to obtain the interaction force change rate dF h The human-machine collaborative control environment is built based on the information of the motion speed V of the center point of the robot's terminal moving platform and the motion speed change rate dV.

[0141] b. Construct a Markov decision process (MDP) with variable admittance parameters in the human-robot collaborative assembly process, specifically including the state space, action space and reward function.

[0142] First, determine the state space of the assembly robot, where the current state s is determined by the interaction force F h , interaction force change rate dF h , the velocity V of the center point of the moving platform and the velocity change rate dV.

[0143] Secondly, determine the action space of human-machine collaboration, and select discrete action a according to the trained decision network, where the action space is discretized into four actions: slightly increase the damping value, slightly decrease the damping value, significantly increase the damping value, and significantly decrease the damping value.

[0144] Finally, the reward function is determined, and the instantaneous reward function r is defined as the inverse of the square of the rate of change of the acceleration of the center point of the robot's terminal moving platform obtained by real-time solution.

[0145] c. Given the starting reference damping value, the spatial pose information of the starting point and the target point, set the batch size (Batch_size), learning rate (Learning_rate), number of training (Episode), attenuation factor (Gamma), memory playback unit size (Memory_size) and the number of layers of the neural network of the DQN algorithm.

[0146] d. Based on the current state s, the neural network of the current value is used to obtain the approximate value function Q(s, a; θ), and then the action a is selected according to the Epsilon-Greedy strategy. During the movement, the robot feeds back the instantaneous reward r based on this action, and the system enters the next state s′. The above process is one step;

[0147] e. The experience sample data (s, a, r, s′) of each time step is stored in the experience pool memory replay unit. When training the network, batches of sample data are randomly extracted from the experience pool memory replay unit to update the network parameters of the current value, breaking the correlation between data and solving the problem of non-static distribution. The DQN error function is updated using the gradient descent method, which is specifically expressed as:

[0148] L(θ)=E[(r+γmax a Q(s′,a′;θ)-Q(s,a;θ)) 2 ]

[0149] After that, every N time steps, the parameters of the current value network are copied to the target value network;

[0150] f. According to the MDP process of the variable admittance parameter, finally according to the termination condition, the target value network is iterated until convergence, and the training is completed.

[0151] like Figure 4 As shown, the variable admittance control process based on machine reinforcement learning specifically includes the following steps:

[0152] a. In the outer loop of admittance control, the interaction force F is collected by the force / torque sensor module h Information and data processing interaction force change rate dF h The information, the motion velocity V of the center point of the moving platform at the end of the robot and the motion velocity change rate dV obtained by solving the solution are used as the environmental input items of reinforcement learning;

[0153] b. According to the MDP process of the variable admittance parameter adjustment, the output admittance parameter damping value B is adjusted in real time based on the DQN algorithm, and the parameter value is transmitted to the admittance controller of the system design;

[0154] c. Calculate the cooperative displacement, velocity and acceleration information of the center point of the terminal moving platform under the action of the human-machine interaction force according to the admittance controller;

[0155] d. Optionally, the admittance model expression is:

[0156]

[0157] in, are the Cartesian space target reference trajectory position, velocity and acceleration preset for the robot, are the actual trajectory position, velocity and acceleration of the Cartesian space target after the robot receives the interaction force applied by the operator. is the interaction force information collected by the force / torque sensor module, are the inertia, damping and stiffness diagonal matrices of the collaborative assembly robot admittance model, respectively.

[0158] In the human-machine collaborative assembly working condition of the present invention, the actual motion trajectory of the robot is the target trajectory, and the robot does not return to the initial posture after the operator removes the human-machine interaction force, so the stiffness diagonal matrix can be eliminated, so the admittance model expression here is transformed into:

[0159]

[0160] e. The collaborative displacement, velocity and acceleration information are used as inputs to the inner loop of the position control of the robot control system and transmitted to the TwinCAT3 controller to solve the control instructions of the servo motor and control the assembly robot to complete the movement.

[0161] The invention has low development difficulty and low cost, and meets the portability requirement. The open structure of the invention is clear and reasonable, easy to operate, safe to run, saves space for motion control cards, reduces costs, gives the host computer real-time computing capabilities, adopts modular thinking to enhance the overall packaging of the system, reduces development difficulty, transmits data through network cables, and the system is portable.

[0162] The present invention has high real-time performance and good reliability. Compared with the traditional field bus, the present invention has high real-time performance and good time certainty, the transmitted information is mostly short frame information, the information exchange is frequent, and the fault tolerance is strong and the reliability is good.

[0163] The present invention can realize adaptive control of human-machine collaborative assembly. When the adaptive admittance control algorithm of the present invention is used for human-machine collaborative assembly operations, the system adjusts the admittance parameters in real time according to the interactive force information collected by the force / torque sensor installed at the end and the robot end motion information to enhance the flexibility and comfort of the collaboration, thereby improving the efficiency of collaborative assembly and reducing the labor intensity of workers.

Claims

1. A human-machine collaborative assembly robot control system, comprising a network main control layer (1), characterized in that: The network master control layer (1) is connected to the master-slave drive control layer (2) for communication, the master-slave drive control layer (2) receives and executes control instructions from a PC host computer in the network master control layer (1) and collects and feeds back data information to the execution component layer, the master-slave drive control layer (2) is connected to the execution component layer (3) for communication, the execution component layer (3) completes the overall motion control of the assembly robot, and the execution component layer (3) feeds back data information; The master-slave drive control layer (2) comprises a communication coupling module (21), the communication coupling module (21) is connected to an input-output module (22), the input-output module (22) connects external input-output variables to an EtherCAT fieldbus network, the input-output module (22) is connected to a force / torque sensor module (23), a driver module (24), and a control cabinet operation panel module (25), the driver module (24) realizes high-precision positioning movement of a servo motor, receives and analyzes relevant instructions of an external control motor, and sends a control signal to a corresponding motor; The force / torque sensor module (23) is used to obtain force / torque information applied by an operator during human-machine collaboration, and the output end of the force / torque sensor module (23) transmits the force / torque information to the input-output module (22).

2. A human-machine collaborative assembly robot control system according to claim 1, characterized in that: The network main control layer (1) includes a human-machine interaction software interface module (12) for realizing human-machine interface interaction. The human-machine interaction software interface module (12) is connected to a PLC main control system module (11). The PLC main control system module (11) realizes communication connection configuration and data interaction of system hardware, and performs real-time control of the hardware. The PLC main control system module (11) is connected to a collaborative motion control algorithm module (13).

3. The human-machine collaborative assembly robot control system according to claim 1, characterized in that: The PC host computer in the network master control layer (1) is an EtherCAT master station, and the master-slave drive control layer (2) is an EtherCAT slave station.

4. The human-machine collaborative assembly robot control system according to claim 1, characterized in that: The driver module (24) is used for high-precision positioning movement of the servo motor. The input port of the driver module (24) receives and analyzes relevant instructions for externally controlling the motor, and the output of the driver module (24) sends a control signal to a corresponding driver (26).

5. The human-machine collaborative assembly robot control system according to claim 1, characterized in that: The execution component layer (3) comprises a motor (31), the motor (31) receives control information from a driver (26) and executes, and the motor (31) is connected to a servo motor module (32).

6. A control method for a human-machine collaborative assembly robot control system according to any one of claims 1 to 5, characterized in that: The following steps are involved: (i) The human-machine interaction software interface module and the control cabinet operation panel module receive the operator's control command information; (ii) the force / torque sensor module receives cooperative control instructions from the operator; (iii) Receive control command information and collaborative control command, and calculate the target motion information of the robot terminal moving platform according to the corresponding algorithm; (vi) Obtain the rod length of each branch chain of the robot through the inverse solution algorithm of the robot mechanism kinematics, and convert it to the encoder disk value of the corresponding servo motor according to the lead pitch relationship of the ball screw; (v) The driver module receives the target value of the encoder and sends a control signal to the corresponding motor according to the currently selected control mode; (vi) The robot's end moving platform is dragged to the vicinity of the target position by the operator through direct contact. According to the actual working conditions, the human-computer interaction software interface module and the control cabinet operation panel module are used to further accurately adjust the end moving platform's posture until the assembly operation is completed; (vii) During collaborative assembly, the operator may press the emergency stop protection button contained in the human-machine interaction software interface module and the control cabinet operation panel module at any time depending on the specific working conditions, or stop contact with the force / torque sensor module to stop the robot from moving.

7. The control method of a human-machine collaborative assembly robot control system according to claim 6, characterized in that: The PLC main control system module is the ADS server, and the collaborative motion control algorithm module is the ADS client. The ADS client sends an ADS request to the ADS server. During the communication process, the client program stops executing until a response is returned by the ADS server.

8. The control method of a human-machine collaborative assembly robot control system according to claim 6, characterized in that: The collaborative motion control algorithm module includes the joint space trajectory planning control algorithm and the human-machine collaborative adaptive admittance control algorithm; The joint space trajectory planning control algorithm is implemented based on a trajectory planning control algorithm of cubic polynomial interpolation; The human-machine collaborative adaptive admittance control algorithm is implemented based on a variable admittance control algorithm of machine reinforcement learning.

Images