Wheel-legged robot whole-body motion control system and method considering slippage steering

Abstract

The invention discloses a wheel-legged robot whole-body motion control system and method considering slippage steering, and relates to the technical field of wheel-legged machine control. The system comprises an input module used for receiving an expected motion track or a motion instruction from an upper layer planning system; the trajectory optimization module takes the motion trajectory output by the input module as a tracking target, so that the platform can be optimized in real time along an offline trajectory; the tracking control module is used for solving a joint torque through inverse dynamics by adopting a total rigid body dynamic model, realizing motion tracking through a whole body control WBC method, and solving a motion and contact force priority task by using a hierarchical optimization method; and the state estimation module is used for estimating all current state quantities of model predictive control MPC and information required by whole body control WBC through sensor information. By means of the system, high-dynamic stable motion control over the wheel-legged robot under the complex terrain can be achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of wheel-legged machine control, and in particular to a whole-body motion control system and method for a wheel-legged robot taking skid steering into consideration. Background Art

[0002] When wheeled-legged robots perform high-speed motion tasks in complex, unstructured environments (such as urban ruins, construction sites, and off-road terrain), they face the following key technical challenges, which seriously restrict their practical application effectiveness:

[0003] 1) Inherent contradiction between mechanical redundancy and dynamic performance:

[0004] Traditional wheel-legged robot control methods (such as ANYmal, Bobcat, and B1W) rely on external swing joints (such as the hip joint's sideways swing degree of freedom) to generate steering torque, resulting in the need for four drive units (two joint motors + one wheel hub motor + one swing motor) per leg, increasing mechanical complexity. This method raises two core issues: (1) High energy consumption and inefficient drive: The swing joints need to frequently output high torque during high-speed steering, resulting in increased system power consumption and a high risk of triggering protective shutdowns due to motor overload. (2) Stability bottleneck: The coupling effect of the external swing joint drive and the high center of gravity structure significantly increases the drive torque required by the external swing joints during high-speed steering, posing a risk of insufficient torque.

[0005] 2) Lack of dynamic modeling of wheel-ground interaction leads to inaccurate control:

[0006] Existing control methods simplify tires into ideal rolling contact points, ignoring the nonlinear coupling relationship between slip rate and dynamic friction. This results in cumulative trajectory tracking errors. On low-adhesion surfaces such as gravel and mud, traditional models fail to compensate for slip, resulting in large lateral deviations between the actual trajectory and the planned path.

[0007] 3) System instability caused by abnormal movement behavior:

[0008] Existing reinforcement learning (RL) control methods lack physical constraint modeling due to inaccurate modeling of wheel-ground contact characteristics by simulation software, which easily leads to two types of abnormal behaviors in high-speed dynamic scenarios: irrational leg lifting: To compensate for insufficient steering torque, the reinforcement learning (RL) controller triggers sudden leg lifting when turning, resulting in an increase in the instantaneous center of mass offset and causing motion oscillation; control command conflict: The control objectives of the hub motor and the joint motor are not coordinated and optimized (for example, the leg posture is not adjusted synchronously when the wheel speed increases), resulting in a vector conflict between the driving torque and the support reaction force. Summary of the Invention

[0009] The technical problem to be solved by the present invention is how to provide a system and method that can achieve high dynamic and stable motion control of a wheel-legged robot under complex terrain.

[0010] To solve the above technical problems, the technical solution adopted by the present invention is: a whole-body motion control system of a wheel-legged robot considering skid steering, comprising:

[0011] An input module is used to receive the desired motion trajectory or motion instructions from the upper planning system and input the desired motion trajectory or motion instructions to the trajectory optimization module for processing;

[0012] The trajectory optimization module is based on model predictive control (MPC) and adopts switching system model predictive control. It uses the output motion trajectory in the input module as the tracking target, enabling the platform to perform real-time optimization along the offline trajectory.

[0013] The tracking control module uses a full rigid body dynamics model to solve the joint torque through inverse dynamics, realizes motion tracking through the whole body control (WBC) method, and uses a hierarchical optimization method to solve the motion and contact force priority tasks;

[0014] The state estimation module is used to estimate all current state quantities of the model predictive control MPC and the information required for whole-body control WBC through sensor information.

[0015] Accordingly, the present invention also discloses a whole-body motion control method for a wheel-legged robot taking into account skid steering, comprising the following steps:

[0016] S1: The input module receives the desired motion trajectory or motion instructions from the upper planning system and inputs the desired motion trajectory or motion instructions to the trajectory optimization module for processing;

[0017] S2: After receiving the desired trajectory, the trajectory optimization module based on model predictive control (MPC) uses the planned motion trajectory in the input module as the tracking target, enabling the platform to perform real-time optimization along the offline trajectory.

[0018] S3: The tracking control module based on whole body control (WBC) receives inputs including the state variables required for inverse dynamics. The inverse dynamics solution is used to obtain the joint and wheel torques required to control the robot.

[0019] S4: The desired joint positions and velocities, the desired wheel velocities, and the joint and wheel torques obtained by the trajectory optimization module are sent to the motors as control commands. Hybrid control is used for both the joint and wheel motors.

[0020] S5: The robot system completes the current control after executing the control instruction.

[0021] The beneficial effects of adopting the above technical solution are: the system and method described in the present invention reduce mechanical complexity through differential wheel speed control without outward swing joints, build a wheel-ground interaction model with dynamic slip compensation to improve control accuracy, and design a composite switching system to suppress abnormal leg lifting behavior, introduce speed differential steering constraints in the model predictive control MPC, and realize highly dynamic and stable motion control under complex terrain. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0023] Figure 1 is a principle block diagram of the system according to an embodiment of the present invention;

[0024] Figure 2 is a diagram defining the coordinate system of the skid steer robot in the system according to an embodiment of the present invention;

[0025] Figure 3 It is a control flow chart of the method described in an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The following is a clear and complete description of the technical solutions in the embodiments of the present invention, in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts are within the scope of protection of the present invention.

[0027] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0028] like Figure 1 As shown, the present invention discloses a whole-body motion control system for a wheel-legged robot with skid steering, comprising an input module, a trajectory optimization module based on model predictive control (MPC), a tracking control module based on whole-body control (WBC), and a state estimation module. The following describes these modules in detail:

[0029] 1. Input module:

[0030] The input module is responsible for receiving the desired motion trajectory (joint trajectory, center of mass trajectory) or the operator's motion instructions (target velocity, target angular velocity, etc.) from the upper-level planning system. If the input is a desired motion trajectory, the trajectory is directly passed as the desired state quantity of the model predictive control (MPC) to the trajectory optimization module based on model predictive control (MPC). If the input is an operator's motion instruction, the desired wheel speed is calculated using the speed differential steering kinematic model, and the state is then passed to the trajectory optimization module based on model predictive control (MPC).

[0031] 2. Trajectory optimization module based on model predictive control MPC:

[0032] The MPC-based trajectory optimization module employs switching system model predictive control, using the planned trajectory in the input module as the tracking target. It solves for the optimal plantar reaction force, the position and velocity of each joint, and the speed and acceleration of the wheels in real time over a long timeframe, enabling the platform to optimize along the offline trajectory in real time. This feedback control reoptimizes the offline trajectory from the measured state, even in the presence of discontinuous transitions and unpredictable interference. Due to the high complexity of wheel-legged platforms with legged systems, to ensure algorithm robustness and real-time planning capabilities, the present invention employs a single rigid-body dynamics model to reflect the dynamic relationship between the center of mass and the foot end. By introducing performance indicators that consider leg type and speed differential steering performance, constraints and optimization objectives are constructed to track the desired trajectory while optimizing the joint states and ground contact forces required for underlying control.

[0033] The MPC model in this paper has 24 state variables, describing the robot's center of mass motion, joint positions, and wheel velocities. The model also has 24 input variables, describing the robot's tire contact forces, joint velocities, and wheel accelerations in the world coordinate system. We define the state variable x(t) and the input variable u(t) as:

[0034]

[0035]

[0036] In the state quantity x(t), v T 、ω T 、p T ,θ T are the velocity, angular velocity, position, and attitude angle of the center of mass, respectively. is the position of the joint, is the wheel speed. Among the input variables, is the contact force between the wheel and the ground, is the joint velocity, is the wheel acceleration.

[0037] The equation of motion for this model is as follows:

[0038]

[0039] Where g is the acceleration of gravity in the world coordinate system, m is the total mass of the single rigid body, I is the moment of inertia about the center of mass, R is the rotation matrix of the body relative to the world coordinate system, and T is the matrix that converts the angular velocity of the base to the world coordinate system. is the Jacobian matrix of the rotation of the center of mass relative to the world coordinate system. fi In addition, the present invention introduces speed differential steering kinematic constraints and tire dynamic constraints in the form of kinematic models and tire models.

[0040] 3. Tracking control module based on WBC:

[0041] Motion planning and control of legged robots face numerous challenges. First, legged robots are typically high-dimensional systems with partially redundant degrees of freedom. Second, interaction with the environment is achieved through multiple contact points where varying contact constraints and interaction forces are imposed. Finally, the environment itself is often unknown, and terrain perception may be limited or completely absent. To reduce the design complexity of motion and control algorithms, motion is often described as a set of simpler tasks, such as maintaining balance through body posture control, locomotion through limb motion, and preventing slipping through contact force constraints. A theoretical framework that accommodates this decomposition is the whole body control (WBC) approach combined with hierarchical optimization.

[0042] A legged robot can be modeled as a free-floating base B with legs attached to it, e.g. Figure 2 As shown. The motion of the entire system can be described by a fixed inertial system I. The position vector from the inertial system to the base system is written as The Hamiltonian unit quaternion is used to parameterize the direction of the vehicle body, and the joint angles are superimposed on the vector In the following form, the generalized coordinate vector q and the generalized velocity vector u are written:

[0043]

[0044] Where n u =6+n j , I v IB and B ω IB are the linear velocity and angular velocity of the base with respect to the inertial coordinate system, q IB The unit quaternion that projects the B coordinate system to the I coordinate system.

[0045] WBC-based tracking control employs a full rigid-body dynamics model, solving for joint torques through inverse dynamics. Motion tracking is achieved via a whole-body controller (WBC), which employs a hierarchical optimization formula to solve a series of motion and contact force-prioritized tasks. The present invention introduces wheel dynamics constraints through the WBC approach to improve the response speed of tracking control.

[0046] 4. State Estimation Module

[0047] The accurate acquisition of the platform motion state parameters is the key foundation for achieving coordinated control of the platform dynamics, and is also a key link in achieving real-time observation of the platform dynamics behavior and accurate feedback of the controller design. The present invention adopts a state estimation method of multi-sensor fusion, which mainly focuses on the state quantities required during the operation of the platform controller. The motion generated by the superposition of the leg odometer and the wheel odometer acts on the center of mass, and the relative displacement of the contact position between the wheel and the ground is used to estimate the motion state of the center of mass relative to the ground. The observation and estimation algorithms of each key state parameter are clarified one by one, and the key parameters of dynamic control such as the position and speed of the center of mass in the world coordinate system are estimated, laying the foundation for online trajectory optimization and whole-body control. The state estimation estimates all the current state quantities of the MPC, the center of mass position, center of mass speed, center of mass attitude angle, center of mass angular velocity, joint position and speed, and wheel speed required by the WBC through the robot sensor information.

[0048] The following is an explanation of the terms that appear in the above modules:

[0049] Wheel-ground interaction dynamics: describes the mechanical interaction between the tires and the ground when a wheeled or wheel-legged robot contacts the ground, including the coupling effects of longitudinal force, lateral force, normal force, and dynamic characteristics such as slip rate and sideslip angle.

[0050] Speed differential slip steering: This technique generates steering torque by controlling the speed difference between the left and right wheels (ωL ≠ ωR), leveraging the slip effect between the tires and the ground to achieve steering without relying on mechanical steering joints. The desired motion trajectory is generated by the planning module and represents the ideal motion path of the robot's center of mass and joints over a period of time. This path typically includes time-varying parameters such as center of mass position, velocity, attitude angle, angular velocity, and joint angles.

[0051] Model Predictive Control (MPC): A control method based on rolling optimization that solves the optimal control problem online within a finite time domain to generate a control sequence that satisfies dynamic constraints. Only the first control step is executed and then re-optimized.

[0052] Whole-body control (WBC): A hierarchical optimization framework that coordinates all degrees of freedom (e.g., joints, wheels) of a robot to accomplish complex tasks, prioritizing high-level tasks (e.g., center of mass stability) over low-level tasks (e.g., end-point trajectory tracking).

[0053] State estimation: Through multi-sensor fusion (IMU, wheel and joint encoders, GPS, lidar, camera, etc.), key information such as the robot's position, speed, and wheel-ground contact status can be inferred in real time to provide observations for the controller's closed-loop control.

[0054] like Figure 3 As shown, an embodiment of the present invention further discloses a whole-body motion control method for a wheel-legged robot considering skid steering, comprising the following steps:

[0055] Step 1: The input module can receive two types of command information from the upper layer. When it receives state quantities such as the gait scheduler, center of mass position, and joint desired angles given by the motion planner, it directly sends the state quantities to the MPC-based trajectory optimization module at a low frequency (5-50Hz). When it receives the gait signal and desired speed issued by the user, such as the desired longitudinal speed given by the handle or keyboard, and angular velocity First, a gait scheduler is generated based on the gait signal. The gait scheduler includes whether each leg touches the ground and the contact time. Then, based on the kinematic model, the speed and angular velocity are input to obtain the expected wheel speed. The calculation formula is as follows:

[0056]

[0057] where w l and w r are the left wheel speed and the right wheel speed respectively, B is the robot wheelbase, r w is the wheel radius.

[0058] The desired position is obtained based on the linear velocity, angular velocity and MPC optimized time domain calculation. The calculation method is as follows:

[0059] Reference trajectory x position = current x position + target x speed * MPC rolling time domain

[0060] Reference trajectory y position = current y position + target y speed * MPC rolling time domain

[0061] Reference trajectory y position = default center of mass height

[0062] Reference trajectory yaw angle = current yaw + target yaw_rat * MPC rolling time domain

[0063] Reference trajectory roll and pitch = 0

[0064] The gait scheduler is divided into 16 subsystems based on whether each leg touches the ground, and contains the duration of each subsystem.

[0065] Finally, the desired speed, desired position, default joint angles, gait scheduler and other information are sent to the MPC-based trajectory optimization module at a low frequency (5-50Hz).

[0066] Step 2: After receiving the desired trajectory, the MPC-based trajectory optimization module uses the desired position, velocity, attitude angle, angular velocity, joint angle, and wheel speed as the state variables to be optimized. The wheel ground contact force, joint angular velocity, and wheel acceleration are used as inputs, and the gait scheduler serves as the switching signal for the MPC switching system. Using a switching system MPC, such as the OCS2 solver, the trajectory is optimized at a rolling frequency of 50-200Hz, taking into account the speed differential steering kinematic constraints and the wheel-ground contact dynamic constraints. The resulting input variables are sent to the WBC-based tracking control module for tracking.

[0067] Step 3: The WBC-based tracking control module receives inputs containing all the states required for inverse dynamics. Therefore, the solution is performed using inverse dynamics (calling a dynamics library such as Pinocchio or RBDL, inputting the states and inputs of the MPC, and then solving for the joint and wheel torques). WBC is an optimization-based control framework that coordinates multiple robot tasks (such as base acceleration, leg swinging, and contact force tracking) while satisfying physical constraints (such as vertical no-motion constraints on the contact leg, joint torque limits, friction cones, and dynamic equations). Furthermore, to incorporate the constraints of the wheel dynamics model, the controller uses a hierarchical optimization formulation to solve a series of motion- and contact-force-first tasks. The optimized states and inputs of the MPC serve as variables for constructing these tasks and constraints. The hierarchical optimization is solved by transforming the series of problems into a quadratic programming form. WBC solves the joint and wheel torques required to control the robot, with a control frequency of 300-1000 Hz.

[0068] Step 4: The desired joint positions and velocities after MPC optimization, the desired wheel velocities, and the joint and wheel torques obtained by WBC are sent to the motors as control commands. Both the joints and wheel motors adopt hybrid control. The control strategy is as follows:

[0069] τ=τ ff +k p ·(p des -p)+k d ·(ω des -ω)

[0070] Among them, τ ff is the desired torque solved by WBC, k p and k d are position stiffness and velocity stiffness respectively, p des and ω desare the desired position and speed solved by MPC, and p and ω are the current position and speed of the motor.

[0071] Step 5: After executing the control command, the robot system completes the current control. Then, the state estimation module reads the attitude angle, angular velocity, and angular velocity data obtained by the nine-axis IMU; reads the motor angle and speed data obtained by the motor encoder; reads the point cloud data obtained by the lidar, and obtains the current positioning information through the laser odometry; and obtains the longitude and latitude in the world coordinate system using the GPS and converts them to the coordinates of the station center coordinate system. The current state is estimated using the Kalman filter algorithm. The state estimation module operates at the same frequency as the World Control Center (WBC).

[0072] The method and system described in this application solve the problems of mechanical redundancy, control inaccuracy, abnormal behavior and insufficient terrain adaptation in the existing technology through four innovations: differential slip steering mechanism, dynamic slip compensation model, contact force optimization algorithm and online terrain adaptive system, and realize high-speed, high-precision and high-reliability motion control of wheel-legged robots in complex unstructured environments, providing technological breakthroughs for scenarios such as urban rescue and field exploration.

Claims

1. A whole-body motion control system for a wheel-legged robot considering skid steering, characterized in that include: An input module is used to receive the desired motion trajectory or motion instructions from the upper planning system and input the desired motion trajectory or motion instructions to the trajectory optimization module for processing; The trajectory optimization module is based on model predictive control (MPC) and adopts switching system model predictive control. It uses the output motion trajectory in the input module as the tracking target, enabling the platform to perform real-time optimization along the offline trajectory. The tracking control module uses a full rigid body dynamics model to solve the joint torque through inverse dynamics, realizes motion tracking through the whole body control (WBC) method, and uses a hierarchical optimization method to solve the motion and contact force priority tasks; The state estimation module is used to estimate all current state quantities of the model predictive control MPC and the information required for whole-body control WBC through sensor information.

2. The whole-body motion control system for a wheel-legged robot taking skid steering into consideration according to claim 1, characterized in that: When the input module inputs the expected motion trajectory, it directly transmits the expected trajectory as the expected state quantity of the model predictive control MPC to the trajectory optimization module based on the model predictive control MPC; when the input is the operator's motion instruction, the expected wheel speed is calculated through the speed difference steering kinematic model, and then the state is transmitted to the trajectory optimization module based on the model predictive control MPC.

3. The whole-body motion control system for a wheel-legged robot taking skid steering into consideration according to claim 1, characterized in that: The trajectory optimization module uses a single rigid body dynamics model to reflect the dynamic relationship between the center of mass and the foot end. By introducing performance indicators that consider leg type and speed difference steering, it constructs constraints and optimization goals, tracks the desired motion trajectory, and optimizes the joint state and ground contact force required for the underlying control. The single rigid body dynamics model is set with multiple inputs, which respectively describe the contact force, joint velocity, and wheel acceleration of the robot's tire in the world coordinate system. The state x(t) and input u(t) are defined as follows: In the state quantity x(t), v T 、ω T 、p T ,θ T are the velocity, angular velocity, position, and attitude angle of the center of mass, respectively. is the position of the joint, is the speed of the wheel; In the input variables, is the contact force between the wheel and the ground, is the joint velocity, is the wheel acceleration; The equation of motion for this model is as follows: Among them, g is the acceleration of gravity in the world coordinate system, m is the total mass of the single rigid body, I is the moment of inertia about the center of mass, R is the rotation matrix of the body relative to the world coordinate system, and T is the matrix that converts the angular velocity of the base to the world coordinate system. is the Jacobian matrix of the rotation of the center of mass relative to the world coordinate system, r fi is the wheel position.

4. The whole-body motion control system for a wheel-legged robot taking skid steering into consideration according to claim 1, characterized in that: The tracking control module adopts a full rigid body dynamics model and solves the joint torque through inverse dynamics. Motion tracking is achieved through whole body control (WBC), and a hierarchical optimization formula is used to solve the motion and contact force priority tasks.

5. The whole-body motion control system of a wheel-legged robot taking skid steering into consideration as claimed in claim 1, characterized in that: The state estimation module: Through the state estimation method of multi-sensor fusion, the motion generated by the superposition of leg odometers and wheel odometers is applied to the center of mass based on the state quantities required during the operation of the platform controller. The relative displacement of the contact position between the wheels and the ground is used to estimate the motion state of the center of mass relative to the ground. The observation and estimation algorithms of each key state parameter are determined, and key parameters of the dynamic control of the center of mass in the world coordinate system are estimated. The state estimation estimates all the current state quantities of the MPC, the center of mass position, center of mass velocity, center of mass attitude angle, center of mass angular velocity, joint position and velocity, and wheel velocity required by the WBC through the robot sensor information.

6. A whole-body motion control method for a wheel-legged robot considering skid steering, characterized in that The steps include: S1: The input module receives the desired motion trajectory or motion instructions from the upper planning system and inputs the desired motion trajectory or motion instructions to the trajectory optimization module for processing; S2: After receiving the desired trajectory, the trajectory optimization module based on model predictive control (MPC) uses the planned motion trajectory in the input module as the tracking target, enabling the platform to perform real-time optimization along the offline trajectory. S3: The tracking control module based on whole body control (WBC) receives inputs including the state variables required for inverse dynamics. The inverse dynamics solution is used to obtain the joint and wheel torques required to control the robot. S4: The desired joint positions and velocities, the desired wheel velocities, and the joint and wheel torques obtained by the trajectory optimization module are sent to the motors as control commands. Hybrid control is used for both the joint and wheel motors. S5: The robot system completes the current control after executing the control instruction.

7. The whole-body motion control method of a wheel-legged robot considering skid steering according to claim 6, characterized in that: The step S1 specifically includes the following steps: The input module receives two types of command information from the upper layer. When it receives the gait scheduler, center of mass position, and desired joint angles given by the motion planner, it directly sends the state quantity at a low frequency to the trajectory optimization module based on model predictive control (MPC). When the received value is the user's published desired longitudinal velocity and angular velocity When , a gait scheduler is first generated based on the gait signal. The gait scheduler includes whether each leg touches the ground and the contact time. Then, based on the kinematic model, the speed and angular velocity are input to obtain the expected wheel speed. The calculation formula is as follows: where w l and w r are the left wheel speed and the right wheel speed respectively, B is the robot wheelbase, r w is the wheel radius; The desired position is obtained based on the linear velocity, angular velocity and MPC optimized time domain calculation. The calculation method is as follows: Reference trajectory x position = current x position + target x speed * MPC rolling time domain; Reference trajectory y position = current y position + target y speed * MPC rolling time domain; Reference trajectory y position = default center of mass height; Reference trajectory yaw yaw angle = current yaw + target yaw_rat * MPC rolling time domain; Reference trajectory roll and pitch = 0; The gait scheduler is divided into 16 subsystems based on whether each leg touches the ground. The duration of each subsystem is included in the gait scheduler. Finally, the desired speed, desired position, default joint angles, and gait scheduler are sent at a low frequency to the trajectory optimization module based on model predictive control (MPC).

8. The whole-body motion control method for a wheel-legged robot considering skid steering according to claim 6, characterized in that: The step S2 specifically includes the following steps: After receiving the desired trajectory, the trajectory optimization module based on model predictive control (MPC) takes the desired position, speed, attitude angle, angular velocity, joint angle, and wheel speed as the state quantities to be optimized, the ground contact force of the wheel, the joint angular velocity, and the wheel acceleration as input, and the gait scheduler as the switching signal of the model predictive control (MPC) switching system. Through the switching system MPC, the motion trajectory is optimized by rolling at a frequency of 50-200Hz. During the optimization process, the speed difference steering kinematic constraints and the wheel-ground contact dynamic constraints are considered, and the solved input quantities are sent to the tracking control module based on whole body control (WBC) for tracking.

9. The whole-body motion control method for a wheel-legged robot considering skid steering according to claim 6, wherein: In the step S4: The control strategy is as follows: τ=τ ff +k p ·(p des -p)+k d ·(ω des -ω); Among them, τ ff is the desired torque for whole body control WBC solution, k p and k d are position stiffness and velocity stiffness respectively, p des and ω des are the expected position and speed solved by the model predictive control MPC, and p and ω are the current position and speed of the motor respectively.

10. The whole-body motion control method of a wheel-legged robot considering skid steering according to claim 6, characterized in that: The S5 robot system completes the current control after executing the control instruction: The state estimation reads the attitude angle, angular velocity, and angular velocity data obtained by the nine-axis IMU; reads the motor angle and speed data obtained by the motor encoder; reads the point cloud data obtained by the lidar, and obtains the current positioning information through the laser odometry; the GPS obtains the longitude and latitude in the world coordinate system and converts it into the coordinates of the station center coordinate system, and estimates the current state through the Kalman filter algorithm. The operating frequency of the state estimation module is the same as that of the whole body control WBC.

Images