Semi-physical simulation platform and method for foot-type robot

Abstract

The invention provides a semi-physical simulation platform and method for a foot-type robot. The platform comprises a measurement and control unit, a motion controller, a four-foot robot three-dimensional virtual model and a sensor. The method comprises the following steps: step 1, firstly building a quadruped robot simulation model and a collision detection model; and step 2, based on the platform, verifying a quadruped robot omni-directional motion control method based on the Trot gait and a large quadruped robot robust motion control method based on the simplified inverted pendulum model. According to the application, the process of verifying the performance of the communication interface and the algorithm through the semi-physical simulation verification platform is adopted, simulation verification before control debugging of the actual prototype platform is realized according to the process, and prior support is provided for debugging of the actual prototype platform.

Description

Technical Field

[0001] The present invention belongs to the technical field of motion control of legged robots, and in particular relates to a semi-physical simulation platform and method for legged robots. Background Art

[0002] The robot's motion controller, serving as the control system's middle coordination layer, communicates with multiple device modules, including the decision controller, servo actuators, measurement and control equipment, attitude sensors, and foot-end force sensors. The motion controller reads the desired motion from the decision controller or remote control terminal, as well as the robot's attitude information collected by the attitude sensors. It then plans and controls the robot's trunk position, implementing foot-end trajectory planning and balance control.

[0003] In the field of quadruped robot motion control, since the data communication methods include multiple modules such as CAN, TCP / UDP, 485, 232, WIFI, etc., the motion control algorithm needs to adapt to the corresponding data interface, and the differences in the constructed three-dimensional model objects, the existing industrial robot control system test platform is not designed for quadruped robots and therefore cannot be directly applied. Summary of the Invention

[0004] The present invention aims to solve the problem that the existing industrial robot control system test platform is not suitable for quadruped robots, and proposes a semi-physical simulation platform and method for legged robots.

[0005] The present application provides a semi-physical simulation platform for legged robots, which includes a measurement and control unit, a motion controller, a three-dimensional virtual model of a quadruped robot, and sensors;

[0006] The measurement and control unit includes a measurement and control computer and measurement and control software; the measurement and control software establishes a network connection with the motion controller, sends control instructions to the motion controller, reads back the motion controller parameters, and displays test data in real time and playbacks historical data to achieve human-computer interaction;

[0007] The motion controller is used to solve the motion of the quadruped platform, plan the foot trajectory, and perform closed-loop posture control. The motion controller software includes motion control algorithms and communication interfaces.

[0008] In the 3D model of the quadruped robot, the virtual model torso is equipped with a virtual 1-channel attitude sensor IMU, 4-channel virtual foot-end force sensors, and virtual joint actuators installed on the virtual joints. The sensors and feedback information of the 3D model are mapped to code, and the various modules and state quantities of the 3D model are represented by the code. CAN communication and network port communication interface drivers are added to open the connection interface between the robot's 3D virtual model and the motion controller, which is used to debug the corresponding communication interface of the motion controller and read back and process sensor data.

[0009] The sensor inputs joint angle commands or joint torque commands through CAN communication and outputs joint angle feedback or joint torque feedback through UDP protocol. The plantar force sensor outputs three-dimensional force through UDP protocol, and the attitude sensor IMU outputs XYZ three-axis attitude angle.

[0010] This application also provides a semi-physical simulation method for a legged robot, which is implemented based on the platform provided by this application and includes:

[0011] Step 1: First, build a quadruped robot simulation model and a collision detection model;

[0012] Step 2: Based on the platform, the omnidirectional motion control method of the quadruped robot based on the Trot gait and the robust motion control method of the large quadruped robot based on the simplified inverted pendulum model were verified respectively;

[0013] Step 2 includes:

[0014] Step 21, verifying the omnidirectional motion control method of the quadruped robot based on the Trot gait;

[0015] Step 22: Verify the robust motion control method of a large quadruped robot based on a simplified inverted pendulum model.

[0016] This application uses the actual motion controller to control the simulated three-dimensional model to achieve simulation verification before the actual prototype platform control debugging, opens up the communication interface between the modules, and obtains preliminary test data as a test reference to provide support for the debugging of the prototype platform. At the same time, the prototype platform test data can in turn further improve the three-dimensional model design, making the simulation model closer to the actual platform and achieving a logical closed loop, thereby ensuring the safety of the test while saving a lot of time and cost, laying a solid foundation for the subsequent implementation of advanced test functions for extreme scenarios and fault injection.

[0017] The present application provides a method for building a semi-physical simulation and verification platform for a legged robot motion controller, which specifically includes the construction of a simulation environment, the mapping between three-dimensional model sensors and codes, the opening of various data communication interfaces, and the verification of motion control algorithms.

[0018] This application verifies the communication interface and algorithm performance through a semi-physical simulation verification platform. Based on this process, simulation verification is achieved before the actual prototype platform control debugging, providing prior support for the debugging of the actual prototype platform.

[0019] In order to more clearly illustrate the functional characteristics and structural parameters of the present invention, further description is given below with reference to the accompanying drawings and specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1This is a schematic diagram of the hardware-in-the-loop system principle provided by an embodiment of the present application;

[0021] Figure 2 This is an architecture diagram of a hardware-in-the-loop simulation verification platform provided in an embodiment of the present application;

[0022] Figure 3 This is a diagram showing the connection relationship and data flow of the hardware-in-the-loop system provided in an embodiment of the present application;

[0023] Figure 4 This is a block diagram of the legged robot motion control method provided in an embodiment of the present application. DETAILED DESCRIPTION

[0024] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0025] The hardware components of the hardware-in-the-loop simulation test platform include a main control module, a data acquisition card, and a display screen. The main control module includes a signal processing module and a model simulation module. Graphical programming is performed using LABVIEW software, and the corresponding test ports are configured to output the required control signals. The data acquisition card is responsible for collecting current signals from the motor or voltage signals from the controller or control system. LABVIEW software analyzes the motor current and speed information and plots the required curves. Finally, by analyzing the specific curves, the control performance of the controller or control system is determined.

[0026] The patent of this invention belongs to the field of quadruped robot motion control development, and specifically involves a method for rapid development of motion controllers. It solves the problem of motion controller function debugging relying on the physical platform of the quadruped robot through a virtual three-dimensional model, realizes the parallel development of the motion controller and servo actuator, and accelerates the development progress.

[0027] This application is based on the slow debugging of the quadruped robot motion controller's hardware and software, the output of the early servo actuators, and the slow debugging progress, as well as safety considerations. It intends to build a motion control hardware-in-the-loop semi-physical simulation platform to debug the motion controller's functions and performance, expose various problems with the motion controller in advance, resolve the problem of the motion controller's functional debugging relying on the quadruped platform, and achieve parallel development of the motion controller and servo actuators, accelerating the development progress. The requirements for the motion controller hardware-in-the-loop testing are as follows:

[0028] (1) Complete the motion controller CAN, TCP / UDP, 485, 232, and WIFI communication interface tests;

[0029] (2) Complete the opening of the control command interface between the host computer, motion controller, and 3D model;

[0030] (3) Complete the testing of the motion controller's motion control algorithm and control logic.

[0031] The present application provides a semi-physical simulation platform for legged robots; the platform includes a measurement and control unit, a motion controller, a three-dimensional virtual model of a quadruped robot, and sensors;

[0032] The measurement and control unit includes a measurement and control computer and measurement and control software; the measurement and control software is written in QT (C++) software and runs on Windows 7 / 10. It establishes a network connection with the motion controller, sends control instructions to the motion controller, reads back the motion controller parameters, and displays test data in real time and historical data playback to achieve human-computer interaction.

[0033] The motion controller uses the EC-A3568J industrial host and the RT-Linux real-time operating system as the software development platform for quadruped platform motion solution, foot end trajectory planning, and posture closed-loop control; the motion controller software includes motion control algorithms and communication interfaces.

[0034] Motion controllers are specifically used for:

[0035] Receive motion control instructions from the measurement and control unit, generate the desired actuator angle, displacement, and torque through the robot motion control algorithm, and send them to the Webots virtual environment through the CAN protocol;

[0036] Receive the current actuator angle, displacement, and torque sent from the Webots virtual environment via CAN or UDP protocol, and participate in the closed-loop control of the angle or torque motion control algorithm;

[0037] The robot's sensor information is sent to the measurement and control unit in real time through the TCP protocol.

[0038] In the 3D model of the quadruped robot, the virtual model torso is equipped with a virtual 1-channel attitude sensor IMU, 4-channel virtual foot-end force sensors, and virtual joint actuators installed on the virtual joints. The sensors and feedback information of the 3D model are mapped to code, and the various modules and state quantities of the 3D model are represented by the code. CAN communication and network port communication interface drivers are added to open the connection interface between the robot's 3D virtual model and the motion controller, which is used to debug the corresponding communication interface of the motion controller and read back and process sensor data.

[0039] The sensor inputs joint angle commands or joint torque commands through CAN communication and outputs joint angle feedback or joint torque feedback through UDP protocol. The plantar force sensor outputs three-dimensional force through UDP protocol, and the attitude sensor IMU outputs XYZ three-axis attitude angle.

[0040] This application also provides a semi-physical simulation method for legged robots:

[0041] Step 1: First, build a quadruped robot simulation model and a collision detection model;

[0042] Among them, the quadruped robot simulation model includes the torso, hip joint, thigh joint, calf joint, position sensor, contact sensor, and IMU;

[0043] At the same time, we built a suitable collision detection model and scene model, including steps of different heights, slopes of different angles, and rugged terrain; as well as the external force impact used in the impact resistance test;

[0044] Step 2: Based on the platform, verify the omnidirectional motion control method of the quadruped robot based on the Trot gait and the robust motion control method of the large quadruped robot based on the simplified inverted pendulum model.

[0045] Step 21: Verify the omnidirectional motion control method of the quadruped robot based on the Trot gait.

[0046] To meet the requirements of omnidirectional motion control of a hydraulic quadruped robot, the unidirectional motion of the quadruped robot on flat terrain is first achieved by planning the rhythmic oscillation motion of the foot end.

[0047] Then, the overall motion of the quadruped robot is decoupled into forward and backward motion, left and right motion, and rotational motion, and the three motions are controlled separately.

[0048] Finally, the foot trajectories are superimposed to achieve omnidirectional motion of the quadruped robot on flat terrain;

[0049] The foot support position adjustment and posture control are decoupled from the basic foot trajectory planning. The slope is adapted based on the actual posture information of the robot torso, enabling adaptive adjustment of the quadruped robot on the slope without affecting the basic foot trajectory planning, thereby extending the omnidirectional mobile control method to stable motion on inclined surfaces.

[0050] Step 22: Verify the robust motion control method for a large quadruped robot based on a simplified inverted pendulum model

[0051] To address the shortcomings of large hydraulic quadruped robots, which are insensitive to external force disturbances and unstructured terrain, a Trot gait motion control method based on a simplified inverted pendulum model was proposed. By integrating the robot's trunk posture angle and angular velocity into the foot trajectory planning, the legs can adjust their support position based on posture information during the stance phase, and the swinging legs can update their foothold position based on the robot's velocity information. This improves the quadruped's mobility and enables robust motion, such as rapid adjustment under lateral impact and adaptability to unstructured terrain.

[0052] During the simulation experiment, the forward speed, turning speed, and step frequency are input to the motion controller;

[0053] The motion controller plans the foot trajectory based on the robot's current pitch angle, yaw angle, foot position, and posture information, and then calculates the joint displacement through leg inverse kinematics. The three-dimensional model is driven to move through the joint controller, and the attitude angle, displacement feedback, and pressure feedback of the three-dimensional model are fed back to the motion controller. The adjusted pitch angle, yaw angle, foot position, and posture information are obtained through the center of mass position adjustment strategy, leg length adjustment strategy, and torso posture adjustment strategy to achieve a closed control loop.

[0054] Before executing step 1, this application also requires platform debugging. The platform debugging method includes:

[0055] a. Joint debugging of the motion controller and the 3D virtual model: Without access to the measurement and control system, the control parameters including gait, movement direction and speed are fixed in the motion control program to carry out relevant functional debugging. First, the data communication interface between the motion controller and the robot's 3D model is opened to debug the motion control algorithm function module in the motion controller, and the control effect is demonstrated through the actual movement of the 3D model.

[0056] b. Joint debugging of the measurement and control system and the motion controller: The motion controller receives the relevant control instructions, parses the parameters, and assigns them to the corresponding motion control parameters. At the same time, the relevant status parameters are fed back to the measurement and control end for display through the TCP protocol. The second step of debugging opens up the interface between the measurement and control system and the motion controller.

[0057] c. Joint debugging of the measurement and control system, motion controller and 3D virtual model; running the measurement and control software and 3D model simulation program simultaneously in the measurement and control system computer. The measurement and control computer and motion controller are connected through the network port and CAN port. The specific physical connection is as follows: Figure 3The measurement and control system sends relevant control parameters to the motion controller, which converts them into joint control instructions after solving them and sends them to the controlled object simulation platform through the CAN port. After the controlled object simulation platform receives the instructions, the joint actuator outputs joint position or force control instructions to drive the four-legged platform three-dimensional model, check the movement of the three-dimensional model, and then verify whether the communication interfaces, kinematic solution, dynamic solution and related control logic of the motion controller are correct; solve the dependence of the motion controller on the four-legged platform when debugging the function and performance.

[0058] d. Joint debugging of the measurement and control system, motion controller, and actual quadruped robot. After completing the joint debugging of the measurement and control system, motion controller, and 3D virtual model, the motion controller is applied to the actual quadruped robot for verification. Its data flow, communication method, and control algorithm remain basically unchanged, which can accelerate the development of the quadruped robot and save a lot of time and costs.

[0059] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0060] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A hardware-in-the-loop simulation platform for legged robots, characterized in that: The platform includes a measurement and control unit, a motion controller, a three-dimensional virtual model of a quadruped robot, and sensors; The measurement and control unit includes a measurement and control computer and measurement and control software; the measurement and control software establishes a network connection with the motion controller, sends control instructions to the motion controller, reads back the motion controller parameters, and displays test data in real time and playbacks historical data to achieve human-computer interaction; The motion controller is used to solve the motion of the quadruped platform, plan the foot trajectory, and perform closed-loop posture control. The motion controller software includes motion control algorithms and communication interfaces. In the 3D model of the quadruped robot, the virtual model torso is equipped with a virtual 1-channel attitude sensor IMU, 4-channel virtual foot-end force sensors, and virtual joint actuators installed on the virtual joints. The sensors and feedback information of the 3D model are mapped to code, and the various modules and state quantities of the 3D model are represented by the code. CAN communication and network port communication interface drivers are added to open the connection interface between the robot's 3D virtual model and the motion controller, which is used to debug the corresponding communication interface of the motion controller and read back and process sensor data. The sensor inputs joint angle commands or joint torque commands through CAN communication and outputs joint angle feedback or joint torque feedback through UDP protocol. The plantar force sensor outputs three-dimensional force through UDP protocol, and the attitude sensor IMU outputs XYZ three-axis attitude angle.

2. The platform according to claim 1, characterized in that Motion controllers are specifically used for: Receive motion control instructions from the measurement and control unit, generate the desired actuator angle, displacement, and torque through the robot motion control algorithm, and send them to the Webots virtual environment through the CAN protocol; Receive the current actuator angle, displacement, and torque sent from the Webots virtual environment via CAN or UDP protocol, and participate in the closed-loop control of the angle or torque motion control algorithm; The robot's sensor information is sent to the measurement and control unit in real time through the TCP protocol.

3. A hardware-in-the-loop simulation method for a legged robot, the method being implemented based on any platform of claims 1 to 2, characterized in that: The method comprises: Step 1: First, build a quadruped robot simulation model and a collision detection model; Step 2: Based on the platform, the omnidirectional motion control method of the quadruped robot based on the Trot gait and the robust motion control method of the large quadruped robot based on the simplified inverted pendulum model were verified respectively; Step 2 includes: Step 21, verifying the omnidirectional motion control method of the quadruped robot based on the Trot gait; Step 22: Verify the robust motion control method of a large quadruped robot based on a simplified inverted pendulum model.

4. The method according to claim 3, characterized in that Step 1: First, build a quadruped robot simulation model and a collision detection model. include: Among them, the quadruped robot simulation model includes the torso, hip joint, thigh joint, calf joint, position sensor, contact sensor, and IMU; At the same time, a suitable collision detection model and scene model are built, including steps of different heights, slopes of different angles, rugged terrain; and the external force impact used in the impact resistance test.

5. The method according to claim 3, characterized in that Step 21, verifying the omnidirectional motion control method of the quadruped robot based on the Trot gait, including: To meet the requirements of omnidirectional motion control of a hydraulic quadruped robot, the unidirectional motion of the quadruped robot on flat terrain is first achieved by planning the rhythmic oscillation motion of the foot end. Then, the overall motion of the quadruped robot is decoupled into forward and backward motion, left and right motion, and rotational motion, and the three motions are controlled separately. Finally, the foot trajectories are superimposed to achieve omnidirectional motion of the quadruped robot on flat terrain; The foot support position adjustment and posture control are decoupled from the basic foot trajectory planning, and the slope is adapted based on the actual posture information of the robot torso.

6. The method according to claim 3, characterized in that Step 22, verifying the robust motion control method of a large quadruped robot based on a simplified inverted pendulum model; including: The robot's trunk posture angle and angular velocity are integrated into the foot trajectory planning, so that the legs in the support phase can adjust their support position according to the posture information, and the swinging legs update the landing point position according to the robot's velocity information. During the simulation experiment, the forward speed, turning speed, and step frequency are input to the motion controller; The motion controller plans the foot trajectory based on the robot's current pitch angle, yaw angle, foot position, and posture information, and then calculates the joint displacement through leg inverse kinematics. The three-dimensional model is driven to move through the joint controller, and the attitude angle, displacement feedback, and pressure feedback of the three-dimensional model are fed back to the motion controller. The adjusted pitch angle, yaw angle, foot position, and posture information are obtained through the center of mass position adjustment strategy, leg length adjustment strategy, and torso posture adjustment strategy to achieve a closed control loop.

7. The method according to claim 3, characterized in that Before executing step 1, you need to perform platform debugging. The platform debugging methods include: a. Joint debugging of the motion controller and the 3D virtual model: Without access to the measurement and control system, the control parameters including gait, movement direction and speed are fixed in the motion control program to carry out relevant functional debugging. First, the data communication interface between the motion controller and the robot's 3D model is opened to debug the motion control algorithm function module in the motion controller, and the control effect is demonstrated through the actual movement of the 3D model. b. Joint debugging of the measurement and control system and the motion controller: The motion controller receives the relevant control instructions, analyzes the parameters, and assigns them to the corresponding motion control parameters. At the same time, it feeds back the relevant status parameters to the measurement and control end through the TCP protocol for display; c. Joint debugging of the measurement and control system, motion controller and three-dimensional virtual model; running the measurement and control software and three-dimensional model simulation program simultaneously in the measurement and control system computer, and connecting the measurement and control computer to the motion controller through the network port and the CAN port; the measurement and control system sends relevant control parameters to the motion controller, which converts them into joint control instructions after solving them, and sends them to the controlled object simulation platform through the CAN port; after the controlled object simulation platform receives the instructions, the joint actuator outputs joint position or force control instructions to drive the four-legged platform three-dimensional model, and checks the movement of the three-dimensional model, thereby verifying whether the communication interfaces, kinematic solutions, dynamic solutions and related control logic of the motion controller are correct; and solving the dependence of the motion controller on the four-legged platform when debugging its functions and performance. d. Joint debugging of the measurement and control system, motion controller, and actual quadruped robot; after completing the joint debugging of the measurement and control system, motion controller, and 3D virtual model, the motion controller is applied to the actual quadruped robot for verification.

Images