Multi-modal motion switching method, apparatus and medium for a wheeled robot

Abstract

This application relates to the field of robot control technology, and discloses a method, device, and medium for multimodal motion switching of a wheeled robot. The method includes: acquiring heading command information and motion state information of the wheeled robot; determining the current motion mode of the wheeled robot based on the motion state information; inputting the motion state information into the dynamic model of the current motion mode to generate balance control quantities for the wheels of the wheeled robot based on the motion state information; generating heading control quantities for the handlebars of the wheeled robot based on the heading command information; and performing mode switching on the wheeled robot based on the balance control quantities and heading control quantities, causing the wheeled robot to switch to a single-track mode or the dual-track mode. The embodiments of this application can achieve smooth switching and stable control of the motion modes of a dual-track to single-track wheeled robot.

Description

Technical Field

[0001] This application relates to the field of robot control technology, and in particular to a method, device and medium for multimodal motion switching of a wheeled robot. Background Technology

[0002] Wheeled robots have garnered widespread attention in service, inspection, and personal transportation sectors due to their structural flexibility and adaptability. Dual-track / single-track wheeled robots, capable of switching motion modes, combine the advantages of both single-track and dual-track motion modes, possessing comprehensive operational capabilities in complex scenarios. However, in related technologies, the motion mode switching process of dual-track / single-track wheeled robots suffers from large amplitude of posture fluctuations and poor dynamic stability, making them highly susceptible to instability during the switching phase. Summary of the Invention

[0003] The purpose of this application is to provide a method, device and medium for switching multimodal motion modes of a wheeled robot, which can realize smooth switching and stable control of motion modes of a dual-track to single-track wheeled robot.

[0004] This application provides a method for multimodal motion switching of a wheeled robot, including:

[0005] Acquire heading command information and motion status information of the wheeled robot;

[0006] Based on the motion state information, the current motion mode of the wheeled robot is determined; the motion mode includes single-track mode, transition mode, and double-track mode;

[0007] The motion state information is input into the dynamic model of the current motion mode to generate balance control quantities for the wheels of the wheeled robot based on the motion state information; the dynamic model of the motion mode is derived based on the dynamic model of the wheeled robot.

[0008] Based on the heading instruction information, a heading control quantity is generated for the handlebars of the wheeled robot;

[0009] Based on the balance control quantity and the heading control quantity, the wheeled robot is switched to a single-track mode or a double-track mode.

[0010] In some embodiments, determining the current motion mode of the wheeled robot based on the motion state information includes:

[0011] Based on the motion state information, the state of the wheeled robot is estimated, and the motion state information is compensated based on the state estimation result to obtain motion state compensation information;

[0012] Based on the motion state compensation information, the current motion mode of the wheeled robot is determined.

[0013] In some embodiments, a method for constructing a dynamic model of the wheeled robot includes:

[0014] Based on the connection method of each component of the wheeled robot, the coordinate transformation matrix and angle-angular velocity mapping relationship of each component of the wheeled robot are calculated;

[0015] Based on the coordinate transformation matrix and the angle-angular velocity mapping relationship, the angle-velocity mapping relationship of each component of the wheeled robot is calculated.

[0016] Based on the angle-angular velocity mapping relationship and the angle-velocity mapping relationship, the angle-kinetic energy mapping relationship of each component of the wheeled robot and the angle-potential energy mapping relationship of the wheeled robot are calculated.

[0017] Based on the angle-kinetic energy mapping relationship and the angle-potential energy mapping relationship, the dynamic model of the wheeled robot is solved using the Lagrange method.

[0018] In some embodiments, a method for constructing a dynamic model of the single-track mode includes:

[0019] Based on the handlebar rotation angle and wheel speed of the wheeled robot, the single-track dynamics model of the wheeled robot is simplified and transformed into a first affine nonlinear system.

[0020] The nonlinearity of the first affine nonlinear system is eliminated by using a feedback linearization method, and the control law of the single-track mode is solved based on the first affine nonlinear system after eliminating the nonlinearity.

[0021] Based on the control law of the single-track mode, a dynamic model of the single-track mode is generated.

[0022] In some embodiments, a method for constructing a dynamic model of the dual-track mode includes:

[0023] Based on the handlebar rotation angle of the wheeled robot, the dual-track dynamics model of the wheeled robot is simplified and linearized to obtain a linear system;

[0024] Based on the linear system, an LQR controller is constructed, and the optimal feedback gain of the LQR controller is solved to obtain the control law of the dual-mode.

[0025] Based on the control law of the dual-track mode, a dynamic model of the dual-track mode is generated.

[0026] In some embodiments, a method for constructing a dynamic model of the transition mode includes:

[0027] Based on the handlebar rotation angle and wheel speed of the wheeled robot, the transition dynamics model of the wheeled robot is simplified and transformed into a second affine nonlinear system; the transition dynamics model of the wheeled robot is obtained by deformation of the single-track dynamics model of the wheeled robot.

[0028] The nonlinearity of the second affine nonlinear system is eliminated by using a feedback linearization method, and the control law of the transition mode is solved based on the second affine nonlinear system after eliminating the nonlinearity.

[0029] Based on the control law of the transition mode, a dynamic model of the transition mode is generated.

[0030] In some embodiments, generating a heading control quantity for the handlebars of the wheeled robot based on the heading instruction information includes:

[0031] Based on the heading instruction information, calculate the corresponding desired turning angle;

[0032] Based on the desired steering angle, a heading control value is generated for the handlebars of the wheeled robot.

[0033] In some embodiments, the modal switching of the wheeled robot based on the balance control quantity and the heading control quantity includes:

[0034] The balance control quantity is sent to the wheel motor driver of the wheel-legged robot, so that the wheel motor driver drives the wheel of the wheel-legged robot to rotate at the corresponding wheel speed.

[0035] The heading control quantity is sent to the handlebar motor driver of the wheeled robot, so that the handlebar motor driver drives the handlebar of the wheeled robot to rotate to the corresponding handlebar rotation angle.

[0036] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described multimodal motion switching method for wheeled robots.

[0037] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described multimodal motion switching method for a wheeled robot.

[0038] The beneficial effects of this application are as follows: By introducing a transition mode and pre-constructing kinematic models for single-track mode, transition mode, and dual-track mode respectively, the heading command information and motion state information are obtained in real time, the current motion mode of the wheeled robot is dynamically determined, and a balance control quantity is generated based on the kinematic model corresponding to the current motion mode of the wheeled robot, as well as a heading control quantity is generated based on the heading command information. This effectively alleviates the problem of severe attitude oscillation caused by sudden changes in dynamic parameters, control law mismatch, and sensor data distortion during the switching process, and avoids the phenomenon of control quantity saturation or reverse correction. Thus, the high-speed off-road capability of single-track mode and the low-speed flexibility of dual-track mode are seamlessly inherited, and the motion stability and comprehensive operation capability of wheeled robot in complex scenarios are significantly improved. Attached Figure Description

[0039] Figure 1 This diagram illustrates the application environment of the multimodal motion switching method for wheeled robots provided in this embodiment.

[0040] Figure 2 This is a flowchart of a multimodal motion switching method for a wheeled robot provided in an embodiment of this application.

[0041] Figure 3 This is a schematic diagram of establishing a coordinate system for a wheeled robot provided in an embodiment of this application.

[0042] Figure 4 This is a schematic diagram of the wheeled robot in single-track mode provided in the embodiments of this application.

[0043] Figure 5 This is a schematic diagram of the wheeled robot in dual-track mode provided in the embodiments of this application.

[0044] Figure 6 This is a schematic diagram of the wheeled robot in a transitional mode provided in the embodiments of this application.

[0045] Figure 7 This is a schematic diagram of the pitch angle change of the wheeled robot provided in the embodiments of this application.

[0046] Figure 8 This is a schematic diagram of the rotation angle change of the wheeled robot provided in the embodiments of this application.

[0047] Figure 9 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0049] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and drawings are used to distinguish similar objects and are not used to describe a specific order or sequence.

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0051] The multimodal motion switching method for wheeled robots provided in this application can be executed by a computer device, which can be a terminal device or a server. The terminal device includes, but is not limited to, mobile phones, computers, smart home appliances, vehicle terminals, and aircraft. The server can be a standalone physical server, a server cluster consisting of multiple physical servers, a distributed system, or a cloud server. Furthermore, all information, data, and signals involved in this application's embodiments are authorized by the relevant parties or have been fully authorized by all parties involved, and the collection, use, and processing of related data comply with the relevant laws, regulations, and standards of the relevant countries and regions.

[0052] In traditional wheeled robot applications, the mode switching process faces technical challenges. Specifically, when a wheeled robot switches from single-track mode to dual-track mode, the abrupt structural change causes a step change in dynamic parameters, leading to a mismatch between the gain of the original control law and the state feedback relationship, resulting in severe oscillations in the attitude angle. Simultaneously, there is a delay in actuator response, and the motor torque output cannot keep up with the control commands in time, causing a lag in attitude adjustment. In addition, sensor data may experience brief distortion or filtering delays during the transition phase, leading to state estimation bias and further amplifying control errors. The combination of these factors causes the system to experience control saturation or reverse correction at the moment of switching, disrupting the balance and ultimately causing the robot to become unstable and fall over. It is impossible to simultaneously retain the high-speed environmental adaptability of single-track mode and the low-speed flexibility of dual-track mode. For example, in an urban road environment, a wheeled robot travels at high speed in single-track mode on open roads. When it needs to enter a narrow alley, it needs to switch to dual-track mode to achieve zero-turn radius steering. During this switching process, a sudden change in dynamic parameters causes a step change in the position of the center of mass and the overall moment of inertia. Control law mismatch causes violent oscillations in the pitch angle. Actuator response delay causes the motor torque output to lag behind the control command. Sensor data distortion causes state estimation deviation. The robot's attitude adjustment fails, causing it to tip over in the alley and interrupting the task execution.

[0053] Therefore, if this problem is not solved, wheeled robots will frequently become unstable during mode switching, making it impossible to reliably perform tasks in complex scenarios. This severely restricts their comprehensive operational capabilities, resulting in their inability to inherit the high-speed off-road capability of the single-track mode and the narrow-space turning capability of the dual-track mode in practical applications.

[0054] Based on this, embodiments of this application provide a method, device, and medium for switching multimodal motion modes of a wheeled robot. By acquiring heading commands and motion state information in real time, the current motion mode is dynamically determined and corresponding balance control quantities and heading control quantities are generated. This effectively solves the control mismatch problem caused by step changes in dynamic parameters during mode switching, and achieves smooth switching and stable control of the motion modes of the wheeled robot.

[0055] Figure 1 This diagram illustrates the application environment of the multimodal motion switching method for wheeled robots provided in this embodiment. (See also...) Figure 1 This method is applied to a multimodal motion switching system for wheeled robots. The system includes a terminal 110 and a server 120. The terminal 110 and server 120 are connected via a network. The terminal 110 can be a desktop terminal or a mobile terminal; the mobile terminal can be at least one of a mobile phone, tablet, or laptop. The server 120 can be a standalone server or a server cluster consisting of several servers. The terminal 110 sends heading command information and the wheeled robot's motion state information to the server 120. The server 120 acquires the heading command information and the wheeled robot's motion state information, determines the current motion mode of the wheeled robot based on the motion state information, inputs the motion state information into the dynamic model of the current motion mode, generates balance control quantities for the wheeled robot's wheels based on the motion state information, generates heading control quantities for the wheeled robot's handlebars based on the heading command information, and performs mode switching on the wheeled robot based on the balance control quantities and heading control quantities, enabling the wheeled robot to switch to a single-track mode or a double-track mode. The motion modes include single-track mode, transition mode and double-track mode, and the dynamic model of the motion modes is derived based on the dynamic model of the wheeled robot.

[0056] It should be understood that Figure 1The application scenarios shown are merely examples. In practical applications, the multimodal motion switching method for wheeled robots provided in this application embodiment can also be applied to other scenarios. For example, the above-described multimodal motion switching method for wheeled robots can be directly applied to terminal 110. Terminal 110 is used to acquire heading command information and motion state information of the wheeled robot. Based on the motion state information, it determines the current motion mode of the wheeled robot and inputs the motion state information into the dynamic model of the current motion mode. Based on the motion state information, it generates balance control quantities for the wheels of the wheeled robot and heading control quantities for the handlebars of the wheeled robot based on the heading command information. Based on the balance control quantities and heading control quantities, it performs mode switching on the wheeled robot, enabling the wheeled robot to switch to single-track mode or double-track mode.

[0057] See Figure 2 In one embodiment, a method for switching multimodal motions of a wheeled robot is provided. The execution subject of this method can be a terminal or a server, including but not limited to steps S201 to S205.

[0058] Step S201: Obtain heading command information and motion state information of the wheeled robot.

[0059] Step S202: Based on the motion state information, determine the current motion mode of the wheeled robot.

[0060] Step S203: Input the motion state information into the dynamic model of the current motion mode to generate the balance control quantity for the wheel of the wheeled robot based on the motion state information.

[0061] Step S204: Based on the heading instruction information, generate the heading control quantity for the handlebars of the wheeled robot.

[0062] Step S205: Based on the balance control quantity and the heading control quantity, the wheeled robot is switched to a single-track mode or a double-track mode.

[0063] Motion modes include single-track mode, transition mode, and dual-track mode. Transition mode refers to the intermediate state of a wheeled robot when switching between single-track and dual-track modes; it can be understood as a transitional phase where the robot's structure deforms but is not fully stable. In practical applications, transition mode identification can be achieved based on threshold judgments of key parameters in the motion state information. For example, when the wheel speed is between a preset high-speed threshold and a low-speed threshold, it is determined to be a transition mode. Alternatively, machine learning algorithms can be used to classify and identify sensor data, primarily to buffer the switching process and avoid attitude angle oscillations caused by abrupt changes in dynamic parameters. Determining the current motion mode of a wheeled robot based on motion state information specifically involves real-time monitoring of the robot's pitch angle and velocity values, comparing them with preset mode determination thresholds. As a preferred implementation, when the absolute value of the pitch angle is less than the first threshold and the speed is greater than the second threshold, it is determined to be a single-track mode; when the absolute value of the pitch angle is greater than the third threshold and the speed is less than the fourth threshold, it is determined to be a dual-track mode; otherwise, it is determined to be a transition mode. This is mainly to achieve dynamic adaptation of the motion state and ensure that the control quantity is generated based on accurate state information.

[0064] The dynamic model of the motion mode is derived from the dynamic model of the wheeled robot. The dynamic model of the motion mode can be implemented using the Newton-Euler method, the Lagrange method, or offline system identification techniques. For example, mathematical equations can be established through physical modeling or parameters can be fitted based on historical data. The main purpose is to achieve a strict match between the control law and the current dynamic parameters, and to avoid lag in motor torque output caused by model mismatch.

[0065] In practical applications, the heading control quantity can be generated using a lookup table method or a simple function mapping. For example, the heading instruction information can be directly matched with the preset handlebar rotation angle value, or the desired steering angle can be calculated through a proportional control algorithm. The main purpose is to achieve independent processing of steering requirements and prevent the coupling with the balance control quantity from amplifying the error.

[0066] The multimodal motion switching method for wheeled robots provided in this application aims to solve the problem of severe attitude oscillations and instability caused by abrupt changes in dynamic parameters, control law mismatch, and sensor data distortion during the switching between single-track and dual-track modes. The multimodal motion switching method first acquires real-time heading command information and motion state information to provide accurate basis for mode switching decisions. The motion state information includes inertial state parameters (vehicle attitude angles, angular velocity, and linear acceleration, etc.) and component state parameters (component position and velocity), ensuring that the control process is based on the current actual state rather than historical data, effectively suppressing state estimation errors caused by sensor filtering delays. Based on the motion state information, the executing entity dynamically determines the current motion mode of the wheeled robot, which includes single-track mode, transition mode, and dual-track mode. The transition mode is introduced as a buffer stage, and continuous monitoring of motion state information avoids abrupt changes in dynamic parameters caused by direct jumps from single-track or dual-track modes, thereby reducing severe attitude angle oscillations. Furthermore, inputting motion state information into the dynamic model of the current motion mode accurately reflects the center of mass position and rotational inertia characteristics under the current motion mode. This generates balance control quantities for the wheels, ensuring that the control law strictly matches the real-time dynamic parameters and avoiding lag or reverse correction in motor torque output due to model mismatch. Simultaneously, based on the heading command information, a heading control quantity for the handlebars is independently generated. This process directly calculates the desired steering angle, ensuring that heading control is not disturbed by balance adjustment and guaranteeing precise execution of steering actions in confined spaces. Based on the balance control quantity and heading control quantity, the executor coordinates the synchronous adjustment of wheel speed and handlebar rotation angle, achieving smooth mode switching and enabling the wheeled robot to stably switch to single-track or double-track modes. Therefore, based on the balance control and heading control variables, the wheeled robot is switched between modes, enabling it to switch to single-track mode or double-track mode. This embodiment effectively avoids the control mismatch problem caused by structural abrupt changes by dynamically adapting the real-time state of the wheeled robot and introducing a transition mode as a buffer stage. Thus, during the switching process, it smoothly inherits the high-speed environment adaptability of single-track mode and the low-speed flexibility of double-track mode, significantly improving the comprehensive operation capability of the wheeled robot in complex scenarios.

[0067] As a specific implementation method, the solution of this application is implemented as follows: In complex outdoor terrain scenarios, the wheeled robot responds to the received mode switching command, starting from single-track mode and eventually switching to dual-track mode, or starting from dual-track mode and eventually switching to single-track mode. During the mode switching process, when the motion state information of the wheeled robot shows that the wheel speed of the wheeled robot gradually decreases and the pitch angle change rate increases significantly, the execution subject identifies the current motion mode of the wheeled robot as a transition mode, and inputs the current motion state information into the dynamic model of the transition mode to generate a balance control quantity. At the same time, it calculates the desired steering angle based on the heading command information input by the user to generate a heading control quantity, and finally completes the switch to dual-track mode or single-track mode through the coordinated action of the wheels and handlebars.

[0068] Through the above technical solutions, the embodiments of this application effectively alleviate the problem of severe attitude oscillation caused by sudden changes in dynamic parameters, control law mismatch and sensor data distortion during the switching process by dynamically adapting to the real-time state and introducing a transition mode. This avoids the phenomenon of control quantity saturation or reverse correction, thereby achieving seamless inheritance of the high-speed off-road capability of the single-track mode and the low-speed flexibility of the dual-track mode, and significantly improving the motion stability and comprehensive operation capability of the wheeled robot in complex scenarios.

[0069] In some embodiments described above, this application proposes determining the current motion mode of a wheeled robot based on motion state information. However, during implementation, sensor data is susceptible to external interference or filtering delays during mode switching transitions, leading to distortion of the original motion state information, resulting in state estimation bias and consequently, incorrect motion mode judgment. Ultimately, this triggers severe attitude angle oscillations and system instability during switching. Therefore, this application further proposes a method for determining the current motion mode of a wheeled robot.

[0070] In some embodiments, determining the current motion mode of the wheeled robot based on motion state information includes: performing state estimation on the wheeled robot based on motion state information, compensating the motion state information based on the state estimation result to obtain motion state compensation information; and determining the current motion mode of the wheeled robot based on the motion state compensation information.

[0071] State estimation refers to the process of inferring the true motion state of a wheeled robot by fusing multi-source sensor data with the system dynamic model. It can be achieved using methods such as Kalman filtering, particle filtering, or extended Kalman filtering. The aim is to dynamically suppress noise interference and filtering delay, and avoid the accumulation of instantaneous errors caused by directly relying on the original data.

[0072] Motion state compensation information can be understood as motion state data after state estimation correction. It can be implemented using weighted averaging, predictive correction, or model-based compensation algorithms. The purpose is to specifically correct the data distortion so that the compensated information more accurately reflects the robot's actual state.

[0073] The proposed solution generates a reliable state estimation result by performing state estimation on the motion state information. This result is used to compensate for the original motion state information, resulting in motion state compensation information. Subsequently, the current motion mode is determined based on the motion state compensation information. Because the state estimation integrates the correlation between the system's historical state and real-time observations, it can effectively suppress sensor data distortion, making the motion state compensation information closer to the true state. This ensures the accuracy of motion mode determination and avoids the risk of attitude oscillations and instability caused by state deviations.

[0074] As a specific implementation method, the solution of this application is implemented as follows: The wheeled robot is equipped with an inertial measurement unit and an encoder. The processor uses a Kalman filter to perform state estimation on the sensor data and generates a state estimation result. The state estimation result is weighted and averaged with the original motion state information to obtain motion state compensation information. Based on the compensation information, the current motion mode is determined to be a single-track mode, a transition mode, or a double-track mode by threshold judgment.

[0075] Through the above technical solutions, the embodiments of this application effectively overcome the interference of sensor data distortion on motion mode determination, significantly reduce the probability of mode misjudgment, ensure that the system accurately identifies the current motion state during the switching transition phase, provide a stable basis for subsequent balance control and heading control, and thus avoid severe attitude angle oscillation and system instability.

[0076] In some embodiments described above, a dynamic model for a wheeled robot was proposed to support balance control during mode switching. However, in its implementation, the construction of the dynamic model lacked systematicity and precision. This resulted in the original control law failing to accurately match the state feedback relationship when dynamic parameters underwent a sudden change during mode switching, leading to severe oscillations in the attitude angle and control saturation, ultimately causing the robot to become unstable and fall over. Therefore, this application further proposes a method for constructing a dynamic model for a wheeled robot.

[0077] In some embodiments, a method for constructing a dynamic model of a wheeled robot includes: calculating the coordinate transformation matrix and angle-angular velocity mapping relationship of each component of the wheeled robot based on the connection method of each component; calculating the angle-velocity mapping relationship of each component of the wheeled robot based on the coordinate transformation matrix and the angle-angular velocity mapping relationship; calculating the angle-kinetic energy mapping relationship and the angle-potential energy mapping relationship of each component of the wheeled robot based on the angle-angular velocity mapping relationship and the angle-velocity mapping relationship; and calculating the dynamic model of the wheeled robot using the Lagrange method based on the angle-kinetic energy mapping relationship and the angle-potential energy mapping relationship.

[0078] A coordinate transformation matrix is ​​a mathematical expression that describes the relative positions and orientations of the components of a wheeled robot. It can be implemented using a homogeneous transformation matrix or a rotation matrix, with the aim of accurately characterizing the geometric constraints of the mechanical structure.

[0079] The angle-angular velocity mapping relationship refers to the dynamic correlation function between joint angle and angular velocity, which can be implemented using Jacobi matrix or differential equations, with the aim of capturing motion coupling characteristics in real time.

[0080] The angle-velocity mapping relationship refers to the conversion law between the angle change and the linear velocity and angular velocity of a wheeled robot. It can be realized by kinematic equations or numerical integration methods, with the aim of reducing the impact of sensor data distortion on state estimation.

[0081] The angle-kinetic energy mapping relationship refers to the correspondence between angle and kinetic energy. It can be derived using energy functions or Lagrange equations, and its purpose is to comprehensively describe the kinetic energy conversion process of the system.

[0082] The angle-potential energy mapping relationship refers to the functional mapping between angle and potential energy. It can be realized using a gravitational potential energy model or an elastic potential energy model, with the aim of accurately characterizing the law of potential energy change.

[0083] The Lagrange method refers to a technical framework for deriving equations of motion based on the energy principle. It can be implemented using the Lagrange equation or Hamilton's principle, with the aim of generating dynamic models that adapt to the nonlinearity of multibody systems.

[0084] This application's solution first calculates the coordinate transformation matrix and angle-angular velocity mapping relationship of each component of the wheeled robot based on the connection method of each component, establishing a precise kinematic basis to reflect the geometric constraints between components. Based on this, the angle-velocity mapping relationship is calculated, integrating kinematic parameters to achieve a dynamic correlation between angle changes and velocity. Furthermore, combining the angle-angular velocity and angle-velocity mapping relationships, the angle-kinetic energy mapping relationship and the angle-potential energy mapping relationship are calculated, deeply fusing kinematic information and energy characteristics to capture the conversion law of kinetic and potential energy. Finally, the Lagrangian method is used to calculate the dynamic model based on the energy mapping relationship, thereby generating a model that accurately describes the energy dynamics of the system during mode switching, effectively addressing the challenges brought about by sudden changes in dynamic parameters.

[0085] As a specific implementation method, the solution of this application is implemented as follows: The components of the wheeled robot include a frame, wheels, and handlebars, which are connected by rotary joints. The coordinate transformation matrix is ​​solved using the Denavit-Hartenberg parametric method. The angle-angular velocity mapping relationship is updated in real time through encoder and inertial measurement unit data. The angle-velocity mapping relationship is calculated based on kinematic equations. The angle-kinetic energy mapping relationship and the angle-potential energy mapping relationship are derived through the mass distribution and geometric parameters of each component. The Lagrangian method is applied to the entire system to generate complete dynamic equations for the wheeled robot to support modal switching control.

[0086] In one specific embodiment, see Figure 3 The coordinate system of the wheeled robot was established and analyzed. The wheeled robot was decomposed into five rigid body models: frame, front and rear handlebars, and front and rear wheels, and their respective coordinate systems were established. The origin is at the center of mass of each rigid body. To simplify the kinematic analysis of the wheeled robot, the following reasonable assumptions are made: (1) All components of the wheeled robot are rigid bodies; (2) The motion of the wheeled robot is completed only on a flat road surface; (3) The motion between the wheels and the ground is pure rolling.

[0087] Solve the coordinate transformation matrix based on the connection method between the rigid body models. Derive the angular velocities of various parts of the wheeled robot. The expressions for the coordinate transformation matrices of each component of the wheeled robot are as follows:

[0088] ,

[0089] ,

[0090] ,

[0091] ,

[0092] The expression for the angle-angular velocity mapping relationship of each component of the wheeled robot is as follows:

[0093] ,

[0094] ,

[0095] ,

[0096] ,

[0097] ,

[0098] in, The roll angle of the frame. The rotation angle of the left handlebar relative to the frame. The rolling angle of the left wheel. This refers to the rotation angle of the right handlebar relative to the frame. Let be the rolling angle of the right wheel. The lateral roll velocity of the chassis, Let be the angular velocity of the left handlebar relative to the frame. Let be the rolling angular velocity of the left wheel. Let be the angular velocity of the right handlebar relative to the frame. ω is the rolling angular velocity of the right wheel.

[0099] Based on the coordinate transformation matrix and the angle-angular velocity mapping relationship, the expressions for the angle-velocity mapping relationships of each component of the wheeled robot are obtained as follows:

[0100] ,

[0101] ,

[0102] ,

[0103] ,

[0104] ,

[0105] in, For the frame speed, The speed of the left handlebar relative to the frame. The speed of the left wheel, The speed of the right handlebar relative to the frame. The speed of the right wheel, This is the distance from the center of gravity of the frame to the midpoint of the line connecting the centers of gravity of the two wheels. The distance from the center of gravity of the handlebars to the center of gravity of the wheels. The radius is the wheel radius.

[0106] Based on the angle-angular velocity mapping relationship and the angle-velocity mapping relationship, the expressions for the angle-kinetic energy mapping relationships of each component of the wheeled robot are obtained as follows:

[0107] ,

[0108] ,

[0109] ,

[0110] ,

[0111] ,

[0112] in, For the kinetic energy of the frame, For the kinetic energy of the left handlebar, For the kinetic energy of the left wheel, For the kinetic energy of the right handlebar, The kinetic energy of the right wheel, For frame quality, For handlebar quality, For the mass of the wheel, Let be the moment of inertia of the flywheel about the y-axis. Let be the moment of inertia of the handlebars about their own x-axis. Let be the moment of inertia of the handlebars about their own y-axis. Let be the moment of inertia of the handlebars about their own z-axis. Let be the moment of inertia of the wheel about its own x-axis. Let be the moment of inertia of the wheel about its own y-axis. Let be the moment of inertia of the wheel about its own z-axis;

[0113] In solving the angle-potential energy mapping relationship of the wheeled robot, the ground is chosen as the zero potential energy surface, and the total potential energy of the wheeled robot is the negative of the gravitational potential energy. When the wheeled robot is in single-track mode, the potential energy of the two wheels changes when the vehicle body tilts over. When the wheeled robot is in double-track mode, the potential energy of the two wheels remains constant when the vehicle body tilts over. The expression for the angle-potential energy mapping relationship of the wheeled robot is:

[0114] ,

[0115] Among them, represents the potential energy of the wheeled robot. This is the acceleration due to gravity.

[0116] Based on the angle-kinetic energy mapping relationship and the angle-potential energy mapping relationship, the expression of the dynamic model of the wheeled robot is obtained by solving the Lagrange method as follows:

[0117] ,

[0118] ,

[0119] ,

[0120] in, The robot's roll angle acceleration. The left wheel's angular acceleration, The angular acceleration of the right wheel. For left-wheel drive torque, This is the right wheel drive torque.

[0121] Through the above technical solutions, the embodiments of this application can ensure that the dynamic model of the wheeled robot accurately reflects the physical characteristics of the wheeled robot, adapt to the step change of dynamic parameters at the moment of mode switching, avoid control law mismatch, effectively suppress attitude angle oscillation and control quantity saturation, thereby ensuring the smoothness and stability of the mode switching process.

[0122] In some embodiments described above, a single-track mode dynamic model is proposed to generate wheel balance control quantities. However, the strong nonlinearity of the single-track dynamic model leads to a complex and inefficient control law solution process, making it unsuitable for real-time requirements during mode switching and potentially causing severe attitude angle oscillations and system instability. Therefore, this application further proposes a method for constructing a single-track mode dynamic model.

[0123] In some embodiments, the method for constructing a dynamic model of a single-track mode includes: simplifying the dynamic model of the single-track robot based on the handlebar rotation angle and wheel speed of the wheel-footed robot and converting it into a first affine nonlinear system; using a feedback linearization method to eliminate the nonlinear part in the first affine nonlinear system, and solving the control law of the single-track mode based on the first affine nonlinear system after eliminating the nonlinear part; and generating a dynamic model of the single-track mode based on the control law of the single-track mode.

[0124] The first affine nonlinear system refers to a class of nonlinear systems in which the control input appears in a linear form in the state equation. It can be realized by rewriting the dynamic model in the form that the state derivative is equal to the sum of the product of the nonlinear function and the control input using the state-space representation method. The purpose is to simplify the model structure, focus on key motion parameters such as handlebar rotation angle and wheel speed, and avoid redundant calculations in the full-state model.

[0125] Feedback linearization can be understood as a control strategy that eliminates system nonlinearity through nonlinear state feedback and coordinate transformation. It can be implemented using precise linearization techniques in differential geometry theory. The aim is to transform complex nonlinear systems into linear controllable forms and reduce the mathematical complexity of control law design.

[0126] The control law for single-track mode is specifically based on a stable control strategy designed after linearization of the system. For example, it can be implemented using the pole placement method. Its purpose is to generate accurate balance control quantities to match the real-time motion state of the wheeled robot.

[0127] This application's solution first simplifies the single-track dynamics model by using handlebar rotation angle and wheel speed as core state variables, focusing the high-dimensional dynamic relationship on key parameters, thus transforming it into a first affine nonlinear system. Subsequently, a feedback linearization method is applied to accurately compensate for nonlinear dynamic effects, transforming the system into a linearly controllable form. Based on this linearized model, the control law for the single-track mode is solved, which can quickly generate equilibrium control variables. Finally, the control law is integrated into the dynamics model to generate a real-time usable model. This step-by-step processing mechanism ensures that the control law can be solved efficiently at the moment of mode switching, avoiding attitude oscillations caused by computational delays, thereby improving the stability of the switching process.

[0128] As a preferred embodiment, the solution of this application is implemented as follows: The control system of the wheeled robot includes a central processing unit, which can be specifically an STM32H7 series microcontroller. During mode switching, the microcontroller acquires the handlebar rotation angle and wheel speed signals in real time, executes a model simplification algorithm to transform the single-track dynamic model into a first affine nonlinear system; then, it calls the feedback linearization module, which eliminates system nonlinearity based on a predefined nonlinear compensation function; next, it uses a state feedback controller to solve the control law of the single-track mode; finally, it outputs the control law to the wheel motor driver to generate a balance control quantity.

[0129] In one specific embodiment, see Figure 4 Given that in single-track mode, , , , Substituting these state conditions into the dynamic model of the wheeled robot, we can obtain the expression for the single-track dynamic model of the wheeled robot as follows:

[0130] ,

[0131] With a fixed wheel speed in single-track mode, the dynamics can be simplified to a simplified single-track dynamics model:

[0132] ,

[0133] The single-track mode uses wheel speed and handlebar speed control to make , To convert the simplified single-track dynamics model into an affine nonlinear system, an equivalent input is introduced: , The expression for the first affine nonlinear system is obtained as follows:

[0134] ,

[0135] ,

[0136] ,

[0137] ,

[0138] in, , , , , , , , , , ;

[0139] Set virtual control quantity Substituting this into the solution of the first affine nonlinear system, we obtain the equivalent input:

[0140] ,

[0141] Given that the wheel speed is constant in single-track mode, the wheel speed can be obtained by measuring it with a sensor. Therefore, the control law for single-track mode can be calculated:

[0142] ,

[0143] A dynamic model of a single-track mode can be designed using a control law based on the single-track mode.

[0144] Through the above scheme, the dynamic model of the single-track mode in this application embodiment can efficiently solve the control law, significantly reduce the computation delay, effectively suppress the violent oscillation of the attitude angle during the mode switching process, ensure that the wheeled robot smoothly switches to the single-track mode, and improve the stability and reliability of the switching process.

[0145] In some embodiments described above, a method for constructing a dynamic model of dual-track modes was proposed to improve control stability. However, during its implementation, the dynamic model of dual-track modes was not specifically optimized, resulting in lag in control response during external disturbances or mode switching, leading to severe attitude angle oscillations and instability risks, making it difficult to meet the dynamic balance requirements in complex scenarios. Therefore, this application further proposes a method for constructing a dynamic model of dual-track modes.

[0146] In some embodiments, the method for constructing a dynamic model of the dual-track mode includes: simplifying and linearizing the dual-track dynamic model of the wheeled robot based on the handlebar rotation angle to obtain a linear system; constructing an LQR controller based on the linear system, solving for the optimal feedback gain of the LQR controller to obtain the control law of the dual-track mode; and generating a dynamic model of the dual-track mode based on the control law of the dual-track mode.

[0147] Simplification and linearization refers to making a linear approximation of the nonlinear dynamic model near a specific operating point of the handlebar rotation angle. This can be achieved by using Jacobi matrix calculation or state space linearization methods. The aim is to reduce the complexity of the model, make the model more closely match the actual operating state, and avoid calculation errors caused by nonlinear characteristics.

[0148] The LQR controller is an optimized controller designed based on the linear quadratic regulation principle. It can combine the system state weight matrix and the control input weight matrix to obtain the optimal feedback gain by solving the Riccati equation. The purpose is to balance control performance and energy consumption and improve the system's robustness to external disturbances.

[0149] This application's solution first simplifies and linearizes the dual-track dynamics model based on the handlebar rotation angle, transforming the nonlinear relationship into a linear system, thus providing a precise mathematical basis for controller design. Based on this, an LQR controller is constructed using this linear system. The control law for the dual-track mode is generated by solving for the optimal feedback gain. This process combines system state and control input weight optimization to ensure automatic adjustment of control strength under external disturbances. Finally, a dynamics model of the dual-track mode is generated based on this control law, allowing the model to directly reflect the optimized control strategy and achieving closed-loop coordination between the model and the control law. These steps form a complete model construction chain. Linearization reduces computational complexity, the global optimization characteristics of the LQR controller suppress attitude oscillations, and the matching between the control law and the dynamics model is ensured during model generation, thereby solving the control response lag problem during external disturbances or mode switching.

[0150] As a preferred embodiment, the solution of this application is implemented as follows: During the operation of the wheeled robot, the handlebar rotation angle is collected in real time as an input parameter to linearize the dual-track dynamics model, resulting in a linear system in state-space form. Subsequently, the lqr function in the MATLAB toolbox is used to solve for the optimal feedback gain of the LQR controller based on the preset state weight matrix and control weight matrix. Finally, this feedback gain is embedded into the dynamics model generation process to form a dual-track modal dynamics model consistent with the control law. In this embodiment, the handlebar rotation angle is used as the core input variable directly related to the steering and balance states. The solution process of the LQR controller does not depend on specific hardware parameters; optimization can be completed only by describing the dynamic characteristics of the system.

[0151] In one specific embodiment, see Figure 5 Given the bi-track mode, , , , Substituting these state conditions into the dynamic model of the wheeled robot, we can obtain the expression for the dual-track dynamic model of the wheeled robot as follows:

[0152] ,

[0153] ,

[0154] ,

[0155] in, For left-wheel drive torque, For right wheel drive torque,

[0156] Without considering the course, let , The dual-track dynamics model of the wheeled robot is simplified to obtain the simplified dual-track dynamics model:

[0157]

[0158] ,

[0159] ,

[0160] Linearize the simplified dual-track dynamics model, and let We obtain the following linear system:

[0161] ,

[0162] ,

[0163] Solve the state-space equations for a linear system, letting the state variables... Control quantity ,have to:

[0164] ,

[0165] Assume the linear system is linear and time-invariant. The state-space equation can be simplified to:

[0166] ,

[0167] Design a state feedback controller. This ensures system stability and minimizes the performance metric (cost function), which is expressed as:

[0168] ,

[0169] in, For performance indicators, State weight matrix (positive semidefinite) ), penalty state deviation, : Control weight matrix (positive definite, ), penalty control energy, T represents transpose,

[0170] Select weight matrix and Then through of The optimal feedback gain can be solved directly using a function. :

[0171] ,

[0172] Among them, by solving the optimal feedback gain By substituting the motion state information measured by the sensor into the solution, the output torque is obtained, and finally the stable balance control of the wheeled robot in the dual-track mode is achieved, that is, the control law of the dual-track mode is determined. Finally, based on the control law of the dual-track mode and the simplified dynamic model above, the dynamic model of the dual-track mode is generated.

[0173] Through the above solution, the embodiments of this application effectively suppress the violent oscillation of the attitude angle under external interference or mode switching scenarios, avoid the risk of instability caused by control saturation or reverse correction, significantly improve the dynamic balance stability of the dual-track mode in narrow spaces or inclined roads, and enable the wheeled robot to reliably maintain flexible steering ability in low-speed scenarios.

[0174] In some embodiments described above, a transition mode is proposed to enable wheeled robots to switch between single-track and dual-track motion modes. However, during its implementation, the dynamic model of the transition phase is not accurately constructed, leading to abrupt changes in dynamic parameters and control law mismatch caused by structural abruptness. This results in severe oscillations of the attitude angle, actuator response lag, and sensor data distortion, ultimately causing switching instability. To address this, this application further proposes a method for constructing a dynamic model for the transition mode.

[0175] In some embodiments, the method for constructing a dynamic model of a transition mode includes: simplifying and transforming the transition dynamic model of the wheeled robot into a second affine nonlinear system based on the handlebar rotation angle and wheel speed of the wheeled robot; using a feedback linearization method to eliminate the nonlinear part in the second affine nonlinear system, and solving the control law of the transition mode based on the second affine nonlinear system after eliminating the nonlinear part; and generating a dynamic model of the transition mode based on the control law of the transition mode.

[0176] The transitional dynamics model of the wheeled robot is derived from the single-track dynamics model of the wheeled robot. The transitional dynamics model is a mathematical representation of the continuous changes in the center of mass position and moment of inertia of the wheeled robot during mode switching. It can be derived from the single-track dynamics model through coordinate system translation and inertial parameter interpolation methods to ensure the continuity of motion with the single-track mode.

[0177] The second affine nonlinear system refers to a mathematical form in which the dynamic equations of the transient phase are decoupled into a linear part and a nonlinear disturbance term. It can be constructed using state-space expressions, with the aim of separating controllable linear dynamics from uncontrollable nonlinear disturbances.

[0178] The control law of the transition mode refers to the feedback control rule designed for the linearized system. It can be generated by combining a state observer with a proportional-derivative control structure, with the aim of providing precise attitude adjustment commands.

[0179] This application's solution first simplifies the transition dynamics model obtained from the deformation of the single-track dynamics model by using the handlebar rotation angle and wheel speed as real-time state variables, thus accurately reflecting the continuous changes in the center of mass position and moment of inertia during structural abrupt changes. The simplified model is then transformed into a second affine nonlinear system, explicitly separating the nonlinear components. Based on this, a feedback linearization method is used to process the second affine nonlinear system, eliminating nonlinear disturbance terms through nonlinear compensation, transforming the dynamic behavior into a linear form. The control law for the transition mode is then solved based on the linearized system, ensuring precise matching between the control command and the current motion state. Finally, a dynamics model for the transition mode is generated based on this control law, ensuring close synchronization between the control output and the actuator response. This process, through the precise construction and linearization of the dynamic model, effectively suppresses attitude angle oscillations and improves control response speed, avoiding control mismatch problems caused by step changes in parameters.

[0180] As a specific implementation method, the scheme of this application is implemented as follows: The transition dynamics model of the wheeled robot can be deformed based on the single-track dynamics model by introducing a function for the continuous change of the center of mass position. This function dynamically adjusts the coordinates of the center of mass according to the rotation angle of the handlebars. In the simplification process, the handlebar rotation angle signal is collected by a magnetic encoder installed on the steering shaft, and the wheel speed signal is provided by a Hall sensor built into the hub motor driver. The transformation of the second affine nonlinear system adopts the state variable substitution method, concentrating the nonlinear terms in the control input channel. The feedback linearization method is specifically implemented through coordinate transformation in differential geometry theory, and the nonlinear compensation term is calculated using Lie derivatives. The control law solution of the transition mode adopts a linear quadratic regulator design, and its state feedback gain matrix is ​​obtained by solving the Riccati equation. The finally generated transition mode dynamics model is deployed to the main controller of the wheeled robot for real-time calculation of the balance control quantity.

[0181] In one specific embodiment, see Figure 6 The expression for the transient dynamics model of a wheeled robot, derived from the deformation of the single-track dynamics model, is as follows:

[0182] ,

[0183] The transition dynamics model of the wheeled robot is simplified to obtain the simplified transition dynamics model:

[0184] ,

[0185] The transition mode is controlled by wheel speed and handlebar speed, so that... , In order to convert the simplified transient dynamics model into an affine nonlinear system, it is known that... Given a known quantity, introduce an equivalent input: , Thus, the second affine nonlinear system is obtained:

[0186] ,

[0187] Feedback linearization is used to eliminate the nonlinearity of the second affine nonlinear system, and a virtual control quantity is assumed. Substituting into the second affine nonlinear system, we obtain the equivalent input:

[0188] ,

[0189] use The rate of change was used to replace ,Right now:

[0190] ,

[0191] in, It was a moment ago. The control law for the transition mode is obtained by solving:

[0192] ,

[0193] The dynamic model of the transition mode can be designed using a control law based on the transition mode.

[0194] Through the above technical solutions, the embodiments of this application realize the continuous characterization of dynamic parameters and the precise matching of control laws during the transition phase, effectively suppress the phenomenon of violent oscillation of attitude angle, reduce the response lag time of actuators, and avoid the amplification of control errors caused by sensor data distortion, thereby ensuring the dynamic stability of the wheeled robot during the switching process between single-track and double-track motion modes, and finally achieving a smooth transition of mode switching.

[0195] In some embodiments described above, this application proposes generating handlebar heading control quantities based on heading command information to achieve steering control of wheeled robots. However, in its implementation, directly generating control quantities based on heading command information lacks a precise calculation step for intermediate target quantities. This makes the steering angle response susceptible to sensor data distortion or filtering delay, failing to effectively compensate for changes in motion state. Consequently, during mode switching, this leads to inaccurate handlebar steering, severe attitude angle oscillations, and increased instability risk. Therefore, this application further proposes a method for generating handlebar heading control quantities for wheeled robots.

[0196] In some embodiments, generating a heading control quantity for the handlebars of a wheeled robot based on heading instruction information includes: calculating a corresponding desired steering angle based on the heading instruction information; and generating a heading control quantity for the handlebars of the wheeled robot based on the desired steering angle.

[0197] The desired steering angle refers to the target steering angle that the wheeled robot handlebars need to achieve. It can be realized by using calculation algorithms based on heading command information and motion state compensation information. For example, filtering algorithms can be used to smooth command abrupt changes and predict dynamic responses. The purpose is to transform the original heading command into a clear steering angle target and avoid direct control deviations caused by sensor noise interference.

[0198] The heading control signal can be understood as the control signal that drives the handlebar motor. It can be implemented by a control algorithm that converts the desired steering angle into a motor drive command. For example, the output can be dynamically adjusted by a proportional-integral-derivative controller or a fuzzy controller. The purpose is to adapt to the handlebar dynamics and ensure that the motor response accurately matches the steering requirements.

[0199] The proposed solution optimizes steering control logic by first converting heading command information into a desired steering angle as an intermediate variable, and then generating a heading control quantity based on this intermediate variable. The heading command information is processed to calculate the desired steering angle. This process integrates motion state compensation information, thereby smoothing sudden command changes and predicting dynamic responses, avoiding steering deviations caused by transient distortions in sensor data. Subsequently, the desired steering angle is used as a stable reference input to the control quantity generation stage, enabling the control quantity generation process to adapt to the handlebar dynamics, dynamically adjusting the motor response speed and amplitude, avoiding steering lag or overshoot caused by actuator delays, ensuring precise matching of handlebar rotation to mode switching requirements, and ultimately achieving coordinated stability between heading control and balance control.

[0200] As a preferred embodiment, the solution of this application is specifically implemented as follows: the processing unit is a microcontroller with an ARM Cortex-M7 architecture, the calculation module uses an extended Kalman filter to fuse heading command information and motion state compensation information to calculate the desired steering angle; the control module uses a digital proportional-integral-derivative controller to convert the desired steering angle into a pulse width modulation signal as a heading control quantity, and drives the handlebar motor to perform steering action through the handlebar motor driver, wherein the microcontroller interacts with the sensor and driver through a serial peripheral interface.

[0201] Through the above-mentioned solution, the embodiments of this application achieve precise steering control and dynamic adaptability during the mode switching process of the wheeled robot, effectively reducing the inaccurate steering of the handlebars and the violent oscillation of the attitude angle caused by sensor data distortion or actuator delay, and significantly improving the system stability and reliability during the mode switching stage.

[0202] In some embodiments described above in this application, during the mode switching process of a wheeled robot based on balance control and heading control, there is a delay in actuator response, and the motor torque output cannot keep up with the control commands in time, resulting in a lag in attitude adjustment. This leads to severe attitude oscillations or instability in the system at the moment of switching. To address this, this application further proposes a method for mode switching of a wheeled robot.

[0203] In some embodiments, the mode switching of the wheeled robot is performed based on the balance control quantity and the heading control quantity, including: sending the balance control quantity to the wheel motor driver of the wheeled robot, so that the wheel motor driver drives the wheel of the wheeled robot to rotate at the corresponding wheel speed; and sending the heading control quantity to the handlebar motor driver of the wheeled robot, so that the handlebar motor driver drives the handlebar of the wheeled robot to rotate to the corresponding handlebar rotation angle.

[0204] The proposed solution directly transmits the balance control signal to the wheel motor driver, enabling the driver to adjust the wheel speed in real time based on this control signal, thus rapidly responding to attitude changes. Simultaneously, the heading control signal is directly transmitted to the handlebar motor driver, allowing the driver to precisely control the handlebar rotation angle and ensure timely execution of steering actions. This direct transmission mechanism eliminates intermediate signal processing steps, avoids processing delays in these steps, guarantees the immediate transmission and execution of balance and heading control commands, and effectively suppresses attitude oscillations caused by actuator response delays.

[0205] As a preferred embodiment, the solution of this application is specifically implemented as follows: The wheel motor driver adopts the L6470 motor driver chip of STMicroelectronics. After receiving the balance control signal, the chip directly drives the hub motor to make the wheel rotate at the target wheel speed; The handlebar motor driver adopts the ASD-AB series servo driver of Nidec Corporation. After receiving the heading control signal, the driver drives the servo motor to rotate the handlebar to the specified angle to ensure the precise execution of the steering action.

[0206] Through the above-described solution, the embodiments of this application effectively reduce the attitude adjustment lag during the mode switching process, suppress the violent attitude oscillation at the moment of switching, and improve the stability and reliability of mode switching of the wheeled robot.

[0207] The multimodal motion switching method for wheeled robots provided in the above embodiments is used to perform modal switching on the wheeled robot. (See also...) Figures 7 to 8 During the transition from dual-track mode to single-track mode, the multi-modal motion switching method for wheeled robots provided in this application is significantly superior to traditional methods. During the dual-track mode phase from 0s to 2s, both control methods can maintain system equilibrium, but during the deformation transition phase from 2s to 4s, Because the control system failed to consider the nonlinear dynamic characteristics of the system, the pitch angle fluctuated drastically, with a maximum overshoot reaching [value missing]. And in Severe instability occurred when switching to single-track mode, eventually tending towards The overturned state. The multimodal motion switching method for wheeled robots provided in this application, through precise modeling and dynamic compensation, effectively suppresses posture oscillations, with the maximum fluctuation being only [missing information]. Furthermore, there was no noticeable impact during the switching process, achieving a smooth transition. The steering angle changed smoothly along the predetermined trajectory, verifying the control system's excellent tracking capability during the configuration evolution process. In summary, the multimodal motion switching method for wheeled robots provided in this application not only ensures stable operation in both dual-track and single-track modes but also achieves smooth and safe switching between modes, significantly improving the motion robustness and adaptability of wheeled robots in complex environments.

[0208] Figure 9 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application.

[0209] The following reference Figure 9 To describe an electronic device 900 according to such an embodiment of the present disclosure. Figure 9 The electronic device 900 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments disclosed herein.

[0210] like Figure 9 As shown, the electronic device 900 is presented in the form of a general-purpose computing device. The components of the electronic device 900 may include, but are not limited to: at least one processing unit 910, at least one storage unit 920, a bus 930 connecting different system components (including storage unit 920 and processing unit 910), a display unit 940, etc.

[0211] The storage unit stores program code, which can be executed by the processing unit 910, causing the processing unit 910 to perform the steps described in the section on the multimodal motion switching method for wheeled robots described in this specification, according to various exemplary embodiments of this disclosure.

[0212] Storage unit 920 may include readable media in the form of volatile storage units, such as random access memory (RAM) 9201 and / or cache memory 9202, and may further include read-only memory (ROM) 9203.

[0213] Storage unit 920 may also include a program / utility 9204 having a set (at least one) program module 9205, such program module 9205 including but not limited to: operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.

[0214] Bus 930 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the various bus structures.

[0215] Electronic device 900 can also communicate with one or more external devices 900' (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable a user to interact with electronic device 900, and / or with any device that enables electronic device 900 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 950. Furthermore, electronic device 900 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 960. Network adapter 960 can communicate with other modules of electronic device 900 via bus 930. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 900, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0216] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0217] The multimodal motion switching method, device, and medium for wheeled robots provided in this application introduce a transition mode and pre-construct kinematic models for single-track mode, transition mode, and dual-track mode, respectively. By acquiring heading command information and motion state information in real time, the current motion mode of the wheeled robot is dynamically determined. Based on the kinematic model corresponding to the current motion mode of the wheeled robot, a balance control quantity is generated, and a heading control quantity is generated based on the heading command information. This effectively alleviates the problem of severe attitude oscillation caused by sudden changes in dynamic parameters, control law mismatch, and sensor data distortion during the switching process, and avoids control quantity saturation or reverse correction phenomena. Thus, it achieves seamless inheritance of the high-speed off-road capability of single-track mode and the low-speed flexibility of dual-track mode, significantly improving the motion stability and comprehensive operation capability of wheeled robots in complex scenarios.

[0218] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, or network device, etc.) to execute the methods described above according to the embodiments of this disclosure.

[0219] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0220] Computer-readable storage media may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable storage medium may also be any readable medium other than a readable storage medium that can transmit, propagate, or transfer a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0221] Those skilled in the art will understand that the above modules can be distributed in the device as described in the embodiments, or they can be modified accordingly and placed in one or more devices that are unique to this embodiment. The modules in the above embodiments can be combined into one module, or they can be further divided into multiple sub-modules.

[0222] Exemplary embodiments of this disclosure have been specifically shown and described above. It should be understood that this disclosure is not limited to the detailed structures, arrangements, or implementations described herein; rather, this disclosure is intended to cover various modifications and equivalent arrangements contained within the spirit and scope of the appended claims.

Claims

1. A method for switching multimodal motions in a wheeled robot, characterized in that, include: Acquire heading command information and motion status information of the wheeled robot; Based on the motion state information, the current motion mode of the wheeled robot is determined; the motion mode includes single-track mode, transition mode, and double-track mode; The motion state information is input into the dynamic model of the current motion mode to generate balance control quantities for the wheels of the wheeled robot based on the motion state information; the dynamic model of the motion mode is derived based on the dynamic model of the wheeled robot. Based on the heading instruction information, a heading control quantity is generated for the handlebars of the wheeled robot; Based on the balance control quantity and the heading control quantity, the wheeled robot is switched to a single-track mode or a double-track mode. A method for constructing the dynamic model of the single-track mode includes: Based on the handlebar rotation angle and wheel speed of the wheeled robot, the single-track dynamics model of the wheeled robot is simplified and transformed into a first affine nonlinear system. The nonlinearity of the first affine nonlinear system is eliminated by using a feedback linearization method, and the control law of the single-track mode is solved based on the first affine nonlinear system after eliminating the nonlinearity. Based on the control law of the single-track mode, a dynamic model of the single-track mode is generated; A method for constructing a dynamic model of the dual-track mode includes: Based on the handlebar rotation angle of the wheeled robot, the dual-track dynamics model of the wheeled robot is simplified and linearized to obtain a linear system; Based on the linear system, an LQR controller is constructed, and the optimal feedback gain of the LQR controller is solved to obtain the control law of the dual-mode. Based on the control law of the dual-track mode, a dynamic model of the dual-track mode is generated; A method for constructing a dynamic model of the transition mode includes: Based on the handlebar rotation angle and wheel speed of the wheeled robot, the transition dynamics model of the wheeled robot is simplified and transformed into a second affine nonlinear system; the transition dynamics model of the wheeled robot is obtained by deformation of the single-track dynamics model of the wheeled robot. The nonlinearity of the second affine nonlinear system is eliminated by using a feedback linearization method, and the control law of the transition mode is solved based on the second affine nonlinear system after eliminating the nonlinearity. Based on the control law of the transition mode, a dynamic model of the transition mode is generated; The step of generating a heading control quantity for the handlebars of the wheeled robot based on the heading command information includes: Based on the heading instruction information, calculate the corresponding desired turning angle; Based on the desired steering angle, a heading control value is generated for the handlebars of the wheeled robot.

2. The wheel-legged robot multi-modal motion switching method according to claim 1, characterized in that, Determining the current motion mode of the wheeled robot based on the motion state information includes: Based on the motion state information, the state of the wheeled robot is estimated, and the motion state information is compensated based on the state estimation result to obtain motion state compensation information; Based on the motion state compensation information, the current motion mode of the wheeled robot is determined.

3. The wheel-legged robot multi-modal motion switching method according to claim 1, characterized in that, The method for constructing the dynamic model of the wheeled robot includes: Based on the connection method of each component of the wheeled robot, the coordinate transformation matrix and angle-angular velocity mapping relationship of each component of the wheeled robot are calculated; Based on the coordinate transformation matrix and the angle-angular velocity mapping relationship, the angle-velocity mapping relationship of each component of the wheeled robot is calculated. Based on the angle-angular velocity mapping relationship and the angle-velocity mapping relationship, the angle-kinetic energy mapping relationship of each component of the wheeled robot and the angle-potential energy mapping relationship of the wheeled robot are calculated. Based on the angle-kinetic energy mapping relationship and the angle-potential energy mapping relationship, the dynamic model of the wheeled robot is solved using the Lagrange method.

4. The wheel-legged robot multi-modal motion switching method according to claim 1, characterized in that, The modal switching of the wheeled robot based on the balance control quantity and the heading control quantity includes: The balance control quantity is sent to the wheel motor driver of the wheel-legged robot, so that the wheel motor driver drives the wheel of the wheel-legged robot to rotate at the corresponding wheel speed. The heading control quantity is sent to the handlebar motor driver of the wheeled robot, so that the handlebar motor driver drives the handlebar of the wheeled robot to rotate to the corresponding handlebar rotation angle.

5. An electronic device, comprising: The electronic device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the multimodal motion switching method for wheeled legged robots according to any one of claims 1 to 4.

6. A computer-readable storage medium storing a computer program, the computer program comprising instructions that, when executed by a computer, cause the computer to perform the method of any one of claims 1 to 5. When the computer program is executed by the processor, it implements the multimodal motion switching method for wheeled robots as described in any one of claims 1 to 4.

Images