High-degree-of-freedom double-leg wheel-foot robot device, control device and control method

Abstract

The invention provides a high-degree-of-freedom double-leg wheel-foot robot device, a control device and a control method. The high-degree-of-freedom double-leg wheel-foot robot device comprises a robot base and at least two leg assemblies. The leg assembly comprises a hip joint mechanism and a mechanical wheel foot, one side of the robot base is connected with the mechanical wheel foot through the hip joint mechanism, and the robot body arranged on the other side of the robot base is driven to move through the mechanical wheel foot; wherein the hip joint mechanism comprises a spherical hinge-like structure formed by a plurality of rotating assemblies, and the rotating assemblies are used for driving the mechanical wheel feet to rotate relative to the robot base along a yaw angle, a rolling angle and a pitching angle according to a received control instruction.

Description

Technical Field

[0001] The present application relates to the field of intelligent robots, and in particular to a high-degree-of-freedom two-legged wheeled robot device, a control device and a control method. Background Art

[0002] With the rapid development of artificial intelligence technology, artificial intelligence technology has also injected more and more vivid machine intelligence into industrial production and daily life. In the process of automation and intelligence of human society, social production urgently needs general-purpose robots with high motion performance and operation performance as algorithm carriers to assist humans in completing work. The combination of general artificial intelligence and general-purpose robots will greatly broaden the application field of robots and improve their efficiency and value in industrial production, medical health, life services and other working conditions. The robot world described by Asimov will gradually become a reality.

[0003] Human production and living scenes are highly complex and unstructured. Bipedal robots are similar to humans in form, and their main design goal is to "imitate human bipedal movements, so that robots can imitate, approach, or even surpass humans in speed, stability, flexibility, and energy efficiency." This is a possible solution for future general-purpose robots.

[0004] In recent years, wheeled-legged robots have received extensive research attention. As a new type of mobile platform, wheeled-legged robots combine the advantages of both the high efficiency of wheels and the flexibility of feet. Not only can they move quickly and flexibly on flat terrain, but they can also effectively cross obstacles. Their superior motion performance has the potential to solve important problems such as movement, transportation, and navigation in complex real-world environments (such as home scenes, disability assistance applications, etc.), so they are considered to be research objects with broad application prospects. However, existing wheeled-legged robots usually have only 1 or 2 degrees of freedom at the hip joint, which limits the robot's tasks such as lateral walking and obstacle avoidance sliding. For example, the Ascento of the Swiss Federal Institute of Technology has only 4 degrees of freedom in its whole body, and its legs (2DOFs) can only realize curvilinear motion in the sagittal plane, and cannot move around the sagittal axis, so it is difficult to travel on a laterally inclined slope; for special motion instructions in the horizontal plane (such as moving to the left front), Ascento must achieve them through differential turning, that is, it must produce additional body rotation and change the orientation of the body, so it cannot meet certain special obstacle avoidance and shooting requirements; the Ollie robot uses a five-bar mechanism, and each leg has 3 degrees of freedom. It can realize free motion in the sagittal plane in the feasible domain, and the leg movement is more flexible, but because the motion plane of its legs is orthogonal to the coronal and transverse planes, it still has limitations under more complex motion tasks. Summary of the invention

[0005] The purpose of this application is to provide a high-degree-of-freedom two-legged wheeled robot device, a control device and a control method. Through the innovation of mechanism-control-algorithm, a high-performance control algorithm for the high-degree-of-freedom wheeled robot is developed to realize human-like movements such as sliding and jumping, so as to be applied to logistics and transportation, special operations, life services and other scenarios.

[0006] To achieve the above-mentioned purpose, the present application provides a high-degree-of-freedom two-legged wheeled robot device, which includes a robot base and at least two leg assemblies; the leg assembly includes a hip joint mechanism and a mechanical wheeled foot, one side of the robot base is connected to the mechanical wheeled foot through the hip joint mechanism, and is used to drive the robot body installed on the other side of the robot base to move through the mechanical wheeled foot; wherein the hip joint mechanism includes a ball-joint structure composed of multiple rotating components, and the rotating component is used to drive the mechanical wheeled foot to rotate relative to the robot base along the yaw angle, roll angle and pitch angle according to the received control instructions.

[0007] In the above-mentioned high-degree-of-freedom two-legged wheeled robot device, optionally, the hip joint mechanism includes a yaw link, a roll link, a thigh link, a first drive motor, a second drive motor and a third drive motor; the fixed side of the first drive motor is arranged on the robot base, and the rotating side is connected to one side of the yaw link, and is used to drive the yaw link to rotate along the yaw angle of the robot base according to the received control command; the fixed side of the second drive motor is arranged on the other side of the yaw link, and the rotating side is connected to one side of the roll link, and is used to drive the roll link to rotate along the roll angle of the robot base according to the received control command; the fixed side of the third drive motor is arranged on the other side of the roll link, and the rotating side is connected to one side of the thigh link, and is used to drive the thigh link to rotate along the pitch angle of the robot base according to the received control command.

[0008] In the above-mentioned high-degree-of-freedom two-legged wheeled foot robot device, optionally, the mechanical wheeled foot includes a calf connecting rod, a foot end wheel, a fourth drive motor and a fifth drive motor; the fixed side of the fourth drive motor is arranged on the other side of the thigh connecting rod, and the rotating side is connected to one side of the calf connecting rod, and is used to drive the calf connecting rod to rotate along the pitch angle of the robot base through a parallel mechanism according to the received control command; the fixed side of the fifth drive motor is arranged on the other side of the calf connecting rod, and the rotating side is connected to the fixed axis of the foot end wheel, and is used to drive the foot end wheel to rotate clockwise or counterclockwise according to the received control command.

[0009] In the above-mentioned high-degree-of-freedom two-legged wheeled robot device, optionally, the fourth drive motor and the thigh link form a parallelogram linkage mechanism; the first drive motor, the second drive motor, the third drive motor, the fourth drive motor and the fifth drive motor all include dual encoders, and the dual encoders are used to receive control instructions provided from the outside, and to feedback the position, speed and torque of the corresponding drive motor.

[0010] The present application also provides a control device applied to the high-degree-of-freedom two-legged wheeled robot device, the control device comprising an inertial measurement module and a microcontroller module; the inertial measurement module is used to collect the operating status of the high-degree-of-freedom two-legged wheeled robot device; the microcontroller is used to provide a communication channel between an external device and a motor through a hybrid communication method; wherein the microcontroller comprises a first microcontroller and a second microcontroller; the first microcontroller is communicatively connected to the first drive motor, the second drive motor and the fourth drive motor through a controller local area network bus, and the second microcontroller is communicatively connected to the third drive motor and the fifth drive motor through a controller local area network bus; the first microcontroller, the second microcontroller and the external device communicate through Ethernet used for controlling automation technology, and the first microcontroller and the second microcontroller are connected in parallel for data transmission.

[0011] The present application also provides a control method applied to the high-degree-of-freedom two-legged wheeled robot device, the method comprising: obtaining control requirements, and obtaining an action flow according to the analysis of the control requirements; obtaining one or more control actions according to the action flow, and obtaining control parameters by calculating the control parameters through the optimal control algorithm and reinforcement learning method according to the action type of the control action; and controlling the high-degree-of-freedom two-legged wheeled robot device according to the control parameters to complete the corresponding action.

[0012] In the above control method, optionally, the action type includes a balancing sliding action and a walking action; when the control action is a balancing sliding action, the control parameters are calculated by an optimal control algorithm; when the control action is a walking action, the control parameters are calculated by a reinforcement learning method; wherein, the control parameters calculated by the optimal control algorithm include: obtaining target data according to the control action analysis, and obtaining a weighting matrix according to the target data by a preset linear quadratic regulator; obtaining the control parameters by a preset control law according to the weighting matrix; obtaining the control parameters by the reinforcement learning method include: obtaining target data according to the control action analysis, and obtaining corresponding control parameters by a target function constructed by a proximal strategy optimization algorithm according to the target data.

[0013] The present application also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above method when executing the computer program.

[0014] The present application also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program for executing the above method.

[0015] The present application also provides a computer program product, comprising a computer program / instruction, which implements the steps of the above method when executed by a processor.

[0016] The beneficial technical effect of the present application is that the bipedal wheeled robot formed by the high-degree-of-freedom bipedal wheeled robot device provided by the present application has three degrees of freedom in each hip, which is similar to humans, so that it has a high degree of flexibility. At the same time, using optimal control theory and reinforcement learning control, a control mode and control method are proposed to enable the robot to perform multimodal motion. Compared with other robots (including legged and wheeled robots) and existing bipedal wheeled robots (such as Ascento of ETH Zurich), this robot has more degrees of freedom in the hip, a more complex structure, and enhanced motion stability and flexibility. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present application, constitute a part of the present application, and do not constitute a limitation of the present application. In the drawings:

[0018] Figure 1A A schematic diagram of the structure of a high-degree-of-freedom two-legged wheeled robot device provided in one embodiment of the present application;

[0019] Figures 1B to 1G A schematic diagram of the structure of each component in a high-degree-of-freedom two-legged wheeled robot device provided in an embodiment of the present application;

[0020] Figure 2 A schematic diagram of the degrees of freedom of a high-degree-of-freedom wheeled-legged robot device provided in one embodiment of the present application;

[0021] Figure 3 A schematic diagram of a control method provided by an embodiment of the present application;

[0022] Figure 4 A schematic diagram of a robot software architecture provided in one embodiment of the present application;

[0023] Figure 5 A schematic diagram of a kinetic model provided in one embodiment of the present application;

[0024] Figure 6A schematic diagram of the structure of an electronic device provided in one embodiment of the present application.

[0025] Reference Numbers

[0026] 100: Robot base

[0027] 101: Robot Hip Joint Motor Yaw Series

[0028] 102: Robot connecting rod Yaw series

[0029] 103: Robot hip joint motor Roll series

[0030] 104: Robot hip joint motor Pitch series

[0031] 105: Robot connecting rod Roll series

[0032] 106: Robotic knee motor

[0033] 107: Robot Thigh Link

[0034] 108: Robot calf connecting rod

[0035] 109: Robot Hub

[0036] 110:Robot wheel motor DETAILED DESCRIPTION

[0037] The following will describe the implementation methods of the present application in detail in conjunction with the accompanying drawings and embodiments, so that the implementation process of how the present application applies technical means to solve technical problems and achieve technical effects can be fully understood and implemented accordingly. It should be noted that as long as there is no conflict, the various embodiments in the present application and the various features in the embodiments can be combined with each other, and the technical solutions formed are all within the protection scope of the present application.

[0038] In addition, the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be performed in an order different from that shown here.

[0039] The present application provides a high-degree-of-freedom two-legged wheeled robot device, which includes a robot base and at least two leg assemblies; the leg assembly includes a hip joint mechanism and a mechanical wheeled foot, one side of the robot base is connected to the mechanical wheeled foot through the hip joint mechanism, and is used to drive the robot body installed on the other side of the robot base to move through the mechanical wheeled foot; wherein the hip joint mechanism includes a ball-joint structure composed of multiple rotating components, and the rotating component is used to drive the mechanical wheeled foot to rotate relative to the robot base along the yaw angle, roll angle and pitch angle according to the received control instructions.

[0040] Specifically, from a biological point of view, the human hip joint has three degrees of freedom to ensure smooth walking. The significant difference between wheeled foot robots and humans is that they use wheels to contact the ground. The structural system of the robot should allow the wheels to have a wide range of postures relative to the ground, thereby providing more types of foot placement methods; for this reason, in the high-degree-of-freedom two-legged wheeled foot robot device provided by this application, the leg assembly is used to simulate the function of two feet. The overall technology is mainly to increase the degree of freedom of the hip joint to improve the stability and flexibility of the wheeled foot robot, and maintain its wheeled and legged movement capabilities. In order to facilitate a clearer understanding of the implementation principle of the degrees of freedom provided by the hip joint mechanism described in this application, the specific structures of the hip joint mechanism and the mechanical wheel foot will be described in detail one by one in the subsequent embodiments, and will not be described in detail here.

[0041] In one embodiment of the present application, the hip joint mechanism includes a yaw link, a roll link, a thigh link, a first drive motor, a second drive motor and a third drive motor; the first drive motor has a fixed side arranged at the robot base, and a rotating side connected to one side of the yaw link, for driving the yaw link to rotate along the yaw angle of the robot base according to a received control instruction; the second drive motor has a fixed side arranged at the other side of the yaw link, and a rotating side connected to one side of the roll link, for driving the roll link to rotate along the roll angle of the robot base according to a received control instruction; the third drive motor has a fixed side arranged at the other side of the roll link, and a rotating side connected to one side of the thigh link, for driving the thigh link to rotate along the pitch angle of the robot base according to a received control instruction.

[0042] In one embodiment of the present application, the mechanical wheel foot includes a calf connecting rod, a foot end wheel, a fourth drive motor and a fifth drive motor; the fixed side of the fourth drive motor is arranged on the other side of the thigh connecting rod, and the rotating side is connected to one side of the calf connecting rod, and is used to drive the calf connecting rod to rotate along the pitch angle of the robot base through a parallel mechanism with the third drive motor according to the received control command; the fixed side of the fifth drive motor is arranged on the other side of the calf connecting rod, and the rotating side is connected to the fixed axis of the foot end wheel, and is used to drive the foot end wheel to rotate clockwise or counterclockwise according to the received control command.

[0043] For details, please refer to Figure 1A As shown, the high-degree-of-freedom two-legged wheeled robot device as a whole includes: a robot base 100, a robot hip joint motor Yaw series 101, a robot connecting rod Yaw series 102, a robot hip joint motor Roll series 103, a robot hip joint motor Pitch series 104, a robot connecting rod Roll series 105, a robot knee motor 106, a robot thigh connecting rod 107, a robot shank connecting rod 108, a robot wheel hub 109 and a robot wheel hub motor 110, wherein, considering that the wheel is in contact with the ground, due to its axial symmetry along the pitch Pitch (along the rotation direction of the wheel), its posture can be described by two coordinates (i.e., the roll angle Roll and the yaw angle Yaw). In the present application, the design of the hip pitch Hip-Pitch and knee pitch Knee-Pitch degrees of freedom can enable the robot to maintain contact with the ground and improve its adaptability to uneven terrain. The wheel posture is achieved by adding hip roll and hip yaw degrees of freedom, while hip pitch and knee pitch ensure walking motion.

[0044] Please refer to Figures 1B to 1GAs shown, with reference to the lower body of a human being, the robot device provided by the present application is composed of a robot base (Base), a yaw link Yaw link of the left and right legs, a roll link Roll link, a thigh link Pitch link, a calf link, and a foot end wheel. This structure enables the robot to have the flexibility of human legs, and at the same time, the wheels are used to replace the human feet, so that the robot can move more efficiently on the ground. Among them, the base plays the role of connecting the two legs, carrying the power supply and the control board, and two motors are installed below, which are connected to the Yaw link; the Yaw link plays the role of connecting the base and the thigh link, which can rotate around the Yaw axis and is fixedly connected to the two motors located at the back of the body and connected to the Roll link; the Roll link plays the role of connecting the base and the thigh link, which can rotate around the Roll and is fixedly connected to the two motors located on the inner side of the body and connected to the thigh link; the thigh link acts as the robot thigh, and a motor is set on the outside to drive the calf link to rotate around the knee joint through a parallel mechanism; the motor is fixed at the end of the calf link to drive the foot end wheel to rotate.

[0045] Please refer to Figure 2 As shown, in the above embodiment, the robot is equipped with joint motors at 10 joints, so that each leg of the robot has 5 degrees of freedom (3 degrees of freedom at the hip joint, 1 degree of freedom at the knee joint, and 1 degree of freedom at the foot end), and the whole machine has 10 degrees of freedom. Taking the left leg as an example, each motor corresponds to 1 degree of freedom, wherein the Hip-Yaw Motor corresponds to the yaw angle of the hip joint, the Hip-Roll Motor corresponds to the roll angle of the hip joint, the Hip-Pitch Motor corresponds to the pitch angle of the hip joint, the Knee Motor corresponds to the pitch angle of the knee joint, and the Wheel Motor corresponds to the rotation angle of the foot. The right leg is similar. The Hip-Yaw Motor, the Hip-Roll Motor, and the Hip-Pitch Motor together form a ball-joint structure, so that each leg can rotate around three axes to simulate the human hip joint. The Knee Motor is connected to the calf through a connecting rod, and together with the thigh, it forms a parallelogram linkage mechanism to drive the calf to rotate, simulating the human knee joint. Wheel Motor can control the robot to move at high speed on the ground, and can realize the rotation of the robot by differentially controlling the rotation speed of the left and right wheels, replacing the human feet with wheels. Therefore, the robot can not only perform simple movements such as moving forward and rotating at high speed, but also realize complex movements such as swinging, squatting, jumping, etc.

[0046] In the above embodiment, the fourth drive motor and the thigh link form a parallelogram linkage mechanism; the first drive motor, the second drive motor, the third drive motor, the fourth drive motor and the fifth drive motor all include dual encoders, which are used to receive control instructions provided externally and to feedback the position, speed and torque of the corresponding drive motor.

[0047] Specifically, the eight motors in the high-degree-of-freedom two-legged wheeled robot device provided in the present application are concentrated near the base. Although the center of gravity height and control difficulty are increased, the weight of the robot legs is reduced, making the leg movements more flexible; in addition, a series-parallel hybrid mechanical connection structure is designed for the legs. Among them, the parallelogram mechanism reduces the weight of the knee joint, reduces the leg inertia, and achieves precise leg control. In addition, it effectively transfers torque from the hip to the knee joint. Therefore, the overall leg structure is kinematically equivalent to a serial manipulator, and this configuration is conducive to kinematic and dynamic solutions; finally, a motor is directly connected to drive the rotation of the wheel. From a practical point of view, hip yaw and hip rolling motion play a vital role in establishing the ground posture of the wheel, while hip pitch motion contributes to the lifting motion of the thigh; knee motion ensures continuous contact between the wheel and the ground to meet the required constraints.

[0048] The present application also provides a control device applied to the high-degree-of-freedom two-legged wheeled robot device, the control device comprising an inertial measurement module and a microcontroller module; the inertial measurement module is used to collect the operating status of the high-degree-of-freedom two-legged wheeled robot device; the microcontroller is used to provide a communication channel between an external device and a motor through a hybrid communication method; wherein the microcontroller comprises a first microcontroller and a second microcontroller; the first microcontroller is communicatively connected to the first drive motor, the second drive motor and the fourth drive motor through a controller local area network bus, and the second microcontroller is communicatively connected to the third drive motor and the fifth drive motor through a controller local area network bus; the first microcontroller, the second microcontroller and the external device communicate through Ethernet used for controlling automation technology, and the first microcontroller and the second microcontroller are connected in parallel for data transmission.

[0049] For details, please refer to Figure 2 As shown, for the unique mechanical structure of the high-degree-of-freedom two-legged wheeled robot device provided by this application, this application provides a corresponding control device for the high-degree-of-freedom two-legged wheeled robot device. In actual work, due to the concentration of motors and aluminum structures at the base and hip, the center of gravity (COG) of the robot is relatively high. The team uses 8 motors to effectively control 8 joints (all joints except the Hip-Pitch Motor) to provide a preset peak torque.

[0050] The peak torque of the Hip-Pitch Motor is selected because it needs to withstand the huge load torque at the knee. For this reason, the peak torque is higher than the peak torque of the above 8 joints. In addition, each motor is equipped with dual encoders, which can accurately feedback the position, speed and torque, and can adapt to various closed-loop control methods. In order to capture the overall state of the wheeled robot, MicroStrain 3DM-GX5-AHRS can be used as an inertial measurement unit (IMU). It has a high-precision accelerometer and a stable gyroscope with a sampling rate of 1000Hz. In terms of communication, two microcontrollers act as a bridge between the motor and the computer. In order to improve real-time performance, a hybrid communication method is adopted between the computer (host), microcontroller (slave) and motor. The host and slave, and the slave and slave communicate through Ethercat (Ethernet for Control Automation Technology). The four slaves are connected in parallel and can transmit data to each other. The host only needs to connect directly to one slave to control all slaves. The slaves and motors communicate via CAN (Controller Area Network), and each slave connects to and controls 2 or 3 motors. The host computer communicates directly with the inertial navigation module (IMU+Gyro) and the game controller via the serial port.

[0051] In terms of power supply, a lithium-ion polymer (LiPo) battery pack can be used to provide stable power for the entire system. After passing through the power module, the output voltage is used to power 10 parallel motors, and the output voltage is converted to supply 4 parallel slave stations at the same time; the host computer uses the user's computer to power the inertial navigation module (IMU+Gyro) and the somatosensory control module; the handle is powered by a built-in power supply, and the power supply is equipped with an air switch to ensure safe use of electricity.

[0052] Please refer to Figure 3 As shown, the present application also provides a control method applied to the high-degree-of-freedom two-legged wheeled robot device, the method comprising:

[0053] S301 obtains control requirements, and obtains an action flow according to the analysis of the control requirements;

[0054] S302 obtains one or more control actions according to the action flow, and obtains control parameters by calculating the optimal control algorithm and the reinforcement learning method according to the action type of the control action;

[0055] S303 controls the high-degree-of-freedom two-legged wheeled robot device to complete corresponding actions according to the control parameters.

[0056] In actual work, in view of the unique mechanical structure of the high-degree-of-freedom two-legged wheeled robot device, this application established a distributed simulation control platform based on IsaacGym+ROS, which mainly consists of three parts: a simulation simulator (Sim) used to import robot models and configuration parameters; a controller (Controller) part that implements the robot's motion control strategy; a communication platform (ROS), which acts as a communication channel to bridge the controller and the simulation platform.

[0057] Experiments can be conducted through the above simulation platform to verify that the wheeled-legged robot has better performance.

[0058] Among them, 1) simulator: simulates the real environment and robot motion. 2) controller: executes the robot motion control algorithm. 3) communicator: connects the controller and simulator through the Robot Operating System (ROS).

[0059] The framework decouples the simulation environment, control algorithm, and communication protocol. This helps to port these algorithms to practical applications. Using IsaacGym as the simulation environment allows large-scale parallelism when training robots on GPUs, and then using PhysX as the physics engine, and deploying LQR and PPO algorithms; based on this framework, multiple environments can be created through APIs to achieve reinforcement learning on GPUs, and the simulation platform can switch traditional control algorithms and conduct reinforcement learning training. At the same time, due to the ROS communication method, the platform has a good communication architecture and debugging API, which can meet the control simulation with high real-time requirements, and can also quickly add new controllers, with strong expansion capabilities. The program part of the simulation environment deployed by the team is mainly composed of three parts running in parallel: RosMaster-a communication layer based on ROS, responsible for multi-threaded information interaction and management; Simulation-a simulation module suitable for Whleaper (the name of this robot) developed based on IsaacGym, responsible for the operation and transformation of the physical simulation environment; and Control Algorithm-a relatively decoupled distributed control algorithm layer that carries the main control strategy.

[0060] Specifically, the control method provided in this application can refer to the software architecture Figure 4As shown in the figure, its control architecture can be divided into three layers: instruction layer, control layer and hardware layer. The instruction layer is responsible for user interaction and command transmission. The user can input the desired posture and speed into the control layer by manipulating the joystick, while observing the real-time feedback of the overall posture and joint status through the user interface (UI). As a key link in feedback control, the control layer is responsible for processing information and issuing commands. It calculates the estimated state vector by integrating data from the IMU and the motor. Subsequently, the specified controller (such as LQR) calculates the motor torque command. These commands are then transmitted to the hardware layer and the robot is smoothly adjusted to the desired state. The hardware layer is responsible for low-level data acquisition and control of hardware components (including IMU and motor). In the underlying control of the motor, a PID controller is used. The hardware layer also forwards the overall posture information from the IMU and the joint status from the motor encoder to the control layer.

[0061] In one embodiment of the present application, the action types include balancing sliding actions and walking actions; when the control action is a balancing sliding action, the control parameters are calculated by an optimal control algorithm; when the control action is a walking action, the control parameters are calculated by a reinforcement learning method; wherein, calculating the control parameters by the optimal control algorithm includes: obtaining target data according to the control action analysis, and obtaining a weighting matrix according to the target data by a preset linear quadratic regulator; obtaining the control parameters by a preset control law according to the weighting matrix; obtaining the control parameters by the reinforcement learning method includes: obtaining target data according to the control action analysis, and obtaining corresponding control parameters by a target function constructed by a proximal strategy optimization algorithm according to the target data.

[0062] In actual work, in order to realize multiple motion modes, namely sliding and walking, the present invention designs a dedicated control algorithm for each motion mode. The optimal control algorithm (LQR) is used to realize balanced sliding, and the reinforcement learning method based on proximal policy optimization (PPO) is used to realize walking, jumping and other actions.

[0063] Specifically, sliding: The balance control given to LQR includes:

[0064] 1) Coordinate system and symbols: Coordinate system such as Figure 5 Established as shown.

[0065] The team represents the LQR state vector as The team denoted the rotation of the hip along the yaw, pitch, and roll axes as γ, β, and α, respectively. The rotation of the knee is denoted by φ, θ represents the rotation of the wheel, while the subscripts r and l represent the right and left sides, respectively. Considering the limited design range of motion of the mechanism, the base, hip, and connected thigh are assumed to have instantaneous constant mass. In addition, considering that the γ and α angles vary within a small range, it is assumed that the leg can be effectively approximated as a planar rigid body in motion. The system is abstracted as a first-order inverted pendulum system (IPS) and established as follows Figure 5 The coordinate system is shown and the symbols are assigned clear labels. Based on the robot's pose, the equivalent mass, inertia, and length matrices can be calculated:

[0066] m,I,l=m,I,l pendulum (γ l ,γ r ,β l ,β r ,α l ,α r ,φ l ,φ r ) (1)

[0067] 2) Derivation of dynamic equations: For a given model, the Lagrangian equation of motion is given by the following equations (2) and (3), where q i Represents each generalized coordinate.

[0068] L=TV (2)

[0069]

[0070]

[0071]

[0072] 3) Stability control: In order to achieve stable sliding motion, a robust stabilization algorithm is required. The goal of this algorithm is to allow the robot to handle external disturbances to maintain balance, using as little time and energy as possible. However, in most cases, saving time and saving energy are contradictory. To solve this problem, the wheeled robot adopts LQR (Linear Quadratic Regulator), which is an optimal control method for controlling linear systems with minimal cost. It has been previously demonstrated that the use of LQR controller can effectively ensure strong reliability and robustness in stability scenarios involving two-wheeled inverted pendulums. The state equation of the LTI (Linear Time Invariant) discrete system is as follows: Equation (6)

[0073] x(k+1)=Ax(k)+Bu(k). (6)

[0074] Where x is the state vector and u is the input vector. The quadratic performance index J is expressed as the discrete time state space representation of the system defined by the control matrices A and B, which is derived from the robot model around the state vector x =

[0075] [0,0,0,0]T is linearized. In the given equation, Q and R represent the weighting matrices of the state covariance, while K represents an unspecified matrix. Equation (7)

[0076]

[0077] Q and R are determined based on the stability required for each state. Experiments have shown that a diagonal weight matrix is ​​sufficient for good control, so the team simplified the model and reduced the number of adjustable weight parameters to the order of the weight matrix. Experiments were conducted in various situations to obtain suitable weight parameters.

[0078] The optimal feedback gain matrix L can be determined by solving the discrete time algebraic Riccati equation (DARE). Equation (8) (9)

[0079] -K+Q+A T KA-A T KB(R+B T AK) -1 B T KA=0 (8)

[0080] L=(R+B T KB) -1 B T KA (9)

[0081] The robot is dynamically adjusted by the following control law. Equation (10)

[0082] u=-Kx (10)

[0083] The state vector x can be derived from the IMU and kinematics. For linear systems, controllability and observability are separable. Therefore, the system can maintain stable operation, resist external disturbances and reach the target position. In addition, for the main posture control of the legs, position control is adopted, which corresponds to the modification of the equivalent mass parameters in the inverted pendulum model.

[0084] Walking: Reinforcement Learning Control:

[0085] To achieve robust walking and jumping control of the 10-DOF robot, the team implemented the PPO algorithm in simulation. PPO (Proximal Policy Optimization) is a reinforcement learning algorithm that ensures stability by limiting the amplitude of policy updates and uses a clipping mechanism to control the extent of the updates. The objective function is given by: Equation (11).

[0086]

[0087] The first term of the minimum function represents L CLIP Item 2 To limit the probability ratio r t (θ) to limit it to between 1-ε and 1+ε. As for rewards, the team used a fine-tuned reward function to regulate the robot's actions, allowing it to walk and jump. The team set the termination and joint position limits to negative values ​​with large absolute values. The goal is to speed up the robot's convergence to a stable state, maintain balance and obey joint constraints. Taking jumping as an example, the team rewards the time its feet leave the ground, the speed along the positive Z axis, and the base height. At the same time, the team penalizes rotation and angular velocity. These lower-value reward functions ensure that the robot's jumps occur in place.

[0088] It is worth noting that in addition to the flexibility of walking and sliding, in the simulation environment, by changing the strategy, the team can freely switch the robot to the sliding mode controlled by LQR or the walking mode controlled by RL. This switching of movement modes enables the robot to adopt different movement modes when dealing with various scenarios, further improving the flexibility of the wheeled robot.

[0089] The beneficial technical effect of the present application is that the bipedal wheeled robot formed by the high-degree-of-freedom bipedal wheeled robot device provided by the present application has three degrees of freedom in each hip, which is similar to humans, so that it has a high degree of flexibility. At the same time, using optimal control theory and reinforcement learning control, a control mode and control method are proposed to enable the robot to perform multimodal motion. Compared with other robots (including legged and wheeled robots) and existing bipedal wheeled robots (such as Ascento of ETH Zurich), this robot has more degrees of freedom in the hip, a more complex structure, and enhanced motion stability and flexibility.

[0090] The present application also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above method when executing the computer program.

[0091] Figure 6 A schematic diagram of the physical structure of an electronic device provided by an embodiment of the present invention, such as Figure 6 As shown, the electronic device includes: a processor (processor) 501, a memory (memory) 502 and a bus 503.

[0092] The processor 501 and the memory 502 communicate with each other via a bus 503 .

[0093] The processor 501 is used to call the program instructions in the memory 502 to execute the methods provided by the above method embodiments.

[0094] An embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program implements the above control method when executed by a processor.

[0095] An embodiment of the present invention further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the control method is implemented.

[0096] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0097] The present invention is described with reference to the flowchart and / or block diagram of the method, device (system), and computer program product according to the embodiment of the present invention. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the process and / or box in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device for implementing the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0098] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device that implements the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0099] These computer program instructions may also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0100] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A high degree of freedom two-legged wheeled robot device, characterized in that: The apparatus comprises a robot base and at least two leg assemblies; The leg assembly includes a hip joint mechanism and a mechanical wheel foot, and one side of the robot base is connected to the mechanical wheel foot through the hip joint mechanism, and is used to drive the robot body installed on the other side of the robot base to move through the mechanical wheel foot; Among them, the hip joint mechanism includes a ball-joint structure composed of multiple rotating components, and the rotating components are used to drive the mechanical wheel foot to rotate along the yaw angle, roll angle and pitch angle relative to the robot base according to the received control instructions.

2. The high-degree-of-freedom two-legged wheeled robot device according to claim 1, characterized in that: The hip joint mechanism comprises a yaw link, a roll link, a thigh link, a first drive motor, a second drive motor and a third drive motor; The first drive motor has a fixed side disposed on the robot base, and a rotating side connected to one side of the yaw link, and is used to drive the yaw link to rotate along the yaw angle of the robot base according to a received control instruction; The second driving motor has a fixed side disposed on the other side of the yaw link, and a rotating side connected to one side of the rolling link, and is used to drive the rolling link to rotate along the rolling angle of the robot base according to the received control command; The fixed side of the third driving motor is arranged at the other side of the rolling link, and the rotating side is connected to one side of the thigh link, and is used to drive the thigh link to rotate along the pitch angle of the robot base according to the received control command.

3. The high-degree-of-freedom two-legged wheeled robot device according to claim 2, characterized in that: The mechanical wheel foot comprises a shank connecting rod, a foot end wheel, a fourth drive motor and a fifth drive motor; The fourth drive motor has a fixed side disposed on the other side of the thigh link, and a rotating side connected to one side of the shank link, and is used to drive the shank link to rotate along the pitch angle of the robot base through a parallel mechanism according to a received control instruction; The fixed side of the fifth driving motor is arranged on the other side of the calf connecting rod, and the rotating side is connected to the fixed axis of the foot end wheel, and is used to drive the foot end wheel to rotate clockwise or counterclockwise according to the received control command.

4. The high-degree-of-freedom two-legged wheeled robot device according to claim 3, characterized in that: The fourth driving motor and the thigh connecting rod form a parallelogram connecting rod mechanism; The first drive motor, the second drive motor, the third drive motor, the fourth drive motor and the fifth drive motor all include dual encoders, and the dual encoders are used to receive control instructions provided externally and to feedback the position, speed and torque of the corresponding drive motor.

5. A control device for the high-degree-of-freedom two-legged wheeled robot device according to claim 3 or 4, characterized in that: The control device comprises an inertial measurement module and a microcontroller module; The inertial measurement module is used to collect the operating status of the high-degree-of-freedom two-legged wheeled robot device; The microcontrol module is used to provide a communication channel between the external device and the motor through a hybrid communication method; Wherein, the microcontroller module includes a first microcontroller and a second microcontroller; The first microcontroller is connected to the first drive motor, the second drive motor and the fourth drive motor via a controller area network bus, and the second microcontroller is connected to the third drive motor and the fifth drive motor via a controller area network bus; The first microcontroller, the second microcontroller and the external device communicate with each other via Ethernet used for controlling automation technology, and the first microcontroller and the second microcontroller are connected in parallel to transmit data.

6. A control method for a high degree of freedom two-legged wheeled robot device according to any one of claims 1 to 5, characterized in that: The method comprises: Obtain control requirements, and obtain action flow according to the analysis of the control requirements; One or more control actions are obtained by disassembling the action flow, and control parameters are obtained by calculating the control parameters through an optimal control algorithm and a reinforcement learning method according to the action type of the control action; The high-degree-of-freedom two-legged wheeled robot device is controlled according to the control parameters to complete corresponding actions.

7. The control method according to claim 6, characterized in that: The action types include balancing sliding action and walking action; When the control action is a balanced sliding action, the control parameters are obtained by calculating the optimal control algorithm; When the control action is a walking action, the control parameters are calculated by a reinforcement learning method; Wherein, obtaining the control parameters by calculating the optimal control algorithm includes: obtaining target data according to the control action analysis, obtaining a weighting matrix according to the target data by calculating the weighting matrix by a preset linear quadratic regulator; obtaining the control parameters by calculating the weighting matrix by a preset control law; Calculating the control parameters by the reinforcement learning method includes: obtaining target data according to the control action analysis, and calculating the corresponding control parameters according to the target data through the objective function constructed by the proximal strategy optimization algorithm.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to claim 6 or 7 is implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to claim 6 or 7 is implemented.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to claim 6 or 7 are implemented.

Images