Control method and device for leg-wheel type robot, leg-wheel type robot and device

Abstract

A control method, device, leg-wheel robot and equipment, which are related to the field of robot control, includes the steps of obtaining (320) current motion state data of the leg-wheel robot, where the current motion state data indicates the motion characteristics of the leg-wheel robot, inputting (340) the current motion state data into a nonlinear controller to obtain a target joint angular acceleration reference value of a target robot joint of the leg-wheel robot, and inputting (360) the target joint angular acceleration reference value into a whole-body dynamics controller to output a joint moment for controlling the leg-wheel robot to perform a control task. The control method combining nonlinear control and whole-body dynamics control not only improves the accuracy of balance control for the leg-wheel robot, but also enhances compliance control for the leg-wheel robot.

Description

[Technical field]

[0001] This application claims priority to a Chinese patent application filed on May 31, 2021, bearing application number 202110604984.4, and entitled "Control method, device, equipment and readable storage medium for leg-wheeled robot," the entire contents of which are incorporated by reference into this application.

[0002] The embodiments of the present application relate to the field of robot control, and in particular to a control method, device, leg-wheel robot, and device for a leg-wheel robot. [Background technology]

[0003] A leg-wheeled robot is a robot structure that controls the motion of the robot body by using a leg-wheel structure. Since a leg-wheeled robot is an unstable underactuated system, it has a problem with balance control. In the related art, a whole-body dynamics controller controls the task execution process of a leg-wheeled robot. Summary of the Invention [Problem to be solved by the invention]

[0004] However, in the related art, it is only possible to control a leg-wheel type robot to complete simple movements. When the target movement is complex, such as jumping, backflips, walking, etc., the distance of the leg-wheel type robot deviates from the balance point significantly and exceeds the linear controllable range, making it impossible to realize balance control for the leg-wheel type robot.

[0005] The problem to be solved is how to realize stable and obedient control of the process of performing complex motion tasks by leg-wheeled robots.

[0006] The embodiments of the present application provide a control method, device, leg-wheeled robot and device, which can improve the stability and accuracy of the control of the leg-wheeled robot. The technical solution is described as follows: [Means for solving the problem]

[0007] According to one aspect, there is provided a method for controlling a leg-wheeled robot executed by a processor of the leg-wheeled robot, the method comprising: acquiring current motion state data of the leg-wheel type robot, the current motion state data indicating motion characteristics of the leg-wheel type robot; Inputting the current motion state data into a nonlinear controller to obtain a target joint angular acceleration reference value of a target robot joint of the leg-wheel robot; and inputting the target joint angular acceleration reference values ​​into a whole-body dynamics controller to output joint moments for controlling the leg-wheeled robot to perform the control task.

[0008] According to another aspect, there is provided a control device for a leg-wheeled robot, the control device comprising: An acquisition module for acquiring current motion state data of the leg-wheel type robot, the current motion state data indicating motion characteristics of the leg-wheel type robot; A decision module inputs the current motion state data into a nonlinear controller to obtain a target joint angular acceleration reference value of a target robot joint of the leg-wheel robot; and an output module that inputs the target joint angular acceleration reference value to the whole body dynamics controller to output joint moments for controlling the leg-wheeled robot to perform the control task.

[0009] According to another aspect, there is provided a leg-wheel type robot, the leg-wheel type robot including a processor and a memory, at least one program stored in the memory, the at least one program being loaded and executed by the processor to realize a control method for the leg-wheel type robot described in any one of the embodiments of the present application above.

[0010] According to another aspect, there is provided a computer device including a processor and a memory, wherein at least one program is stored in the memory, and the at least one program is loaded and executed by the processor to realize a method for controlling a leg-wheeled robot described in any one of the embodiments of the present application above.

[0011] According to another aspect, a computer-readable storage medium is provided, the storage medium having at least one program stored therein, the at least one program being loaded and executed by a processor to realize a method for controlling a leg-wheeled robot as described in any one of the above embodiments of the present application.

[0012] According to another aspect, a chip is provided, the chip including a programmable logic circuit or a program, the chip implementing the method for controlling a leg-wheeled robot according to any one of the above embodiments.

[0013] According to another aspect, there is provided a computer program product, the computer program product including computer instructions stored in a computer readable storage medium, the computer instructions being read by a processor of a computing device from the computer readable storage medium and executed to cause the computing device to perform the method for controlling a leg-wheeled robot according to any one of the above embodiments. Effect of the Invention

[0014] The beneficial effects of the technical solutions provided by the embodiments of the present application at least include: By using the target joint angular acceleration reference value of the target robot joint of the leg-wheeled robot determined by the nonlinear controller as the input of the whole-body dynamics controller, the advantages of the stable control of the robot using the nonlinear system control method and the compliance control of the robot using the whole-body dynamics method can be combined to achieve both the precision of the balance control of the robot and the flexibility of the compliance control of the robot, thereby controlling the leg-wheeled robot to obediently complete complex movements such as jumping, backflips, and walking. [Brief description of the drawings]

[0015] [Figure 1] FIG. 1 is a structural schematic diagram of a leg-wheel type robot provided by one exemplary embodiment of the present application. [Diagram 2] 1A-1C are schematic diagrams of a leg-wheel robot provided by an exemplary embodiment of the present application at different heights; [Diagram 3] 1 is a flowchart of a control method for a leg-wheel robot provided by one exemplary embodiment of the present application. [Figure 4] 1 is a flowchart of a control method for a leg-wheel robot provided by one exemplary embodiment of the present application. [Diagram 5] 1 is a flowchart of a control method for a leg-wheel robot provided by one exemplary embodiment of the present application. [Figure 6] 1 is a flowchart of a method for determining joint moments of each joint when a leg-wheel type robot performs a control task according to an exemplary embodiment of the present application. [Figure 7] FIG. 1 is a three-dimensional schematic diagram of a leg-wheel type robot provided by one exemplary embodiment of the present application. [Figure 8] FIG. 1 is a schematic diagram of a control structure of a leg-wheel robot provided by one exemplary embodiment of the present application. [Figure 9] FIG. 2 is a structural block diagram of a control device for a leg-wheel type robot provided by one exemplary embodiment of the present application. [Figure 10]FIG. 1 is a structural block diagram of a leg-wheel type robot provided by one exemplary embodiment of the present application. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0016] In order to make the objectives, technical solutions and advantages of the present application clearer, the embodiments of the present application are further described in detail below in conjunction with the drawings.

[0017] First, the terms used in the embodiments of the present application will be introduced. Leg-wheel robot: A leg-wheel robot is a robot structure that controls the motion of the robot base by the leg-wheel structure, and since the contact points between the leg-wheel robot and the ground only include the contact points between the wheels and the ground, if the leg-wheel structure arrangement itself is unstable, there is a problem with balance control. The upper part of the leg-wheel structure is the thigh structure including the joints, and the lower part is the drive structure of the wheel structure.

[0018] The leg-wheel robot is any robot that includes a wheel structure, and the specific structure of the leg-wheel robot can be set according to actual needs. For example, the leg-wheel robot is realized as a leg-wheel biped robot, and the leg-wheel part of the leg-wheel biped robot includes two wheels for movement, and the two wheels are respectively connected to the thigh structure of the leg-wheel biped robot and connected to the base part of the robot through the thigh structure, and the two wheels drive the base part to complete the movement control.

[0019] Preferably, the number of thigh structures of the leg-wheel biped robot is four, and every two thigh structures are connected to one wheel, and each thigh structure is connected via a shin segment and a thigh segment via a rotation joint. For example, if the leg-wheel biped robot includes a thigh structure A, a thigh structure B, a thigh structure C, and a thigh structure D, the thigh structure A and the thigh structure B are connected to the first wheel (regarded as the left wheel), and the thigh structure C and the thigh structure D are connected to the second wheel (regarded as the right wheel). The thigh structure A, the thigh structure B and the first wheel, and the thigh structure C, the thigh structure D and the second wheel constitute a two-thigh plane parallel connection structure of the leg-wheel robot. Each parallel connection structure has five rotation joints, and has two translational degrees of freedom in the lateral and vertical directions, respectively.

[0020] Controller: In the embodiments of the present application, it is a pre-established method for performing balance control on the leg-wheeled robot, which is generally expressed in a mathematical formula and implemented by code programming. In some embodiments, the controller performs balance control on the leg-wheeled robot by combining a linearized control method and a non-linearized control method when the leg-wheeled robot is out of the vicinity of the balance point and cannot perform balance control using a linear model. It also performs motion control and compliance control on the leg-wheeled robot.

[0021] The leg-wheel type robot includes a leg-wheel section and a base section connected to the leg-wheel section, and each of the leg-wheel section and the base section includes at least one joint.

[0022] The leg-wheel unit is a wheel-shaped member for realizing the motion of the leg-wheel robot. The leg-wheel unit may be, for example, one wheel, two wheels, four wheels, or any other number of leg-wheel units, and each leg-wheel unit may be controlled, for example, by two thigh units connected in parallel or multiple thigh units connected in series. Here, the present application does not limit the specific structural type of the leg-wheel unit and the number of its wheels.

[0023] The base part is a main body part of the leg-wheel type robot, and may be, for example, a body part of the leg-wheel type robot, and the base part may be, for example, a flat plate-like member connected to the leg-wheel part of the leg-wheel type robot, or may have other shapes according to actual needs. The present application does not limit the specific shape and structure of the base part.

[0024] For example, the base part is connected to the leg wheel part by a holder, or the connection between the base part and the leg wheel part is realized in other manners, and the present application does not limit the specific connection manner between the base part and the leg wheel part.

[0025] In some embodiments, the base unit may include, for example, six joints (three-dimensional coordinate positions and three-dimensional Euler angles) and the leg-wheel unit may include, for example, two joints, or the base unit may include one joint and the leg-wheel unit may include eight joints. The present application does not limit the specific number of joints included in the base unit and the leg-wheel unit, nor does it limit the specific joint configuration of the leg-wheel type robot.

[0026] 1 is a structural schematic diagram of a leg-wheel robot 100 provided by one exemplary embodiment of the present application. As shown in FIG. 1, the leg-wheel robot 100 may be a leg-wheel biped robot, and the leg-wheel robot 100 includes a base part 110 and a leg-wheel part 120; The base unit 110 is connected to the leg-wheel unit 120, which includes two wheels 121 and a thigh structure 122 connecting the wheels 121 and the base unit 110. As shown in FIG. 1, the leg-wheel robot 100 includes a total of four thigh structures 122, and two of the four thigh structures 122 are connected to one wheel 121. In general, when there are thigh structures A, B, C, and D, the thigh structures A and B are connected to the first wheel, and the thigh structures C and D are connected to the second wheel. The thigh structures A and B and the first wheel, and the thigh structures C and D and the second wheel constitute the two-thigh plane parallel connection structure of the leg-wheel robot 100. The parallel connection type thigh has five rotational joints and has two translational degrees of freedom in the lateral and vertical directions. Compared with the serial mechanism, the parallel connection mechanism has the characteristics of compact structure, high rigidity and strong load-bearing capacity, so that the robot can jump higher and overcome obstacles more flexibly.

[0027] Optionally, the thigh structure 122 includes a shin segment 1221 and a thigh segment 1222, the shin segment 1221 and the thigh segment 1222 being connected by a pivot joint, and the shin segment 1221 and the wheel 121 being connected by a pivot joint.

[0028] Four sets of motors are provided on the base 110 to correspond to the four thigh structures 122, and the four sets of motors control the bending and stretching of the thigh structures 122, and in some embodiments, the segments connecting the thigh structures 122 and the base 110 are connected by a rotating joint, and generally, as shown in Fig. 1, when the motor drives the rotating joint to rotate clockwise, the thigh structure 122 is controlled to change to bending, and when the motor drives the rotating joint to rotate counterclockwise, the thigh structure 122 is controlled to change to stretch (the manner in which the two sets of thigh structures 122 are driven by the rotating joints is similar or different). That is, the relationship between the clockwise-counterclockwise rotation manner and the bending-stretching control mode is similar or different.

[0029] The bending and stretching of the thigh structure 122 (i.e., the relative positional relationship between the shin segment 1221 and the thigh segment 1222) is used to control the height of the leg-wheel robot 100, i.e., when the thigh structure 122 changes to bend, the height of the leg-wheel robot 100 decreases, and when the thigh structure 122 changes to stretch, the height of the leg-wheel robot 100 increases. Schematically, referring to FIG. 2, the thigh structure 122 shown in FIG. 1 is a case where the bending degree is large, at this time, the height of the leg-wheel robot 100 is low, and the bending degree of the thigh structure 122 in FIG. 2 is smaller than that of the thigh structure 122 shown in FIG. 1, and at this bending degree, the height of the leg-wheel robot 100 is high, and the balance control moment at the two heights is different because the balance of the leg-wheel robot is different at different heights in FIG. 1 and FIG. 2, so the balance control moment at the two heights is also different.

[0030] The wheel 121 is a driving wheel, that is, after being driven by a motor, the wheel 121 realizes active rotation, and further realizes motion state control for the leg-wheel robot 100, for example, controlling the leg-wheel robot 100 to move forward, backward, turn, or stop.

[0031] Based on the structure of the base part 110 and the leg-wheel part 120 in the leg-wheel robot 100, the leg-wheel robot 100 is made to approximate the structure of a miniature inverted pendulum. The height of the leg-wheel robot 100 corresponds to the pendulum length in the inverted pendulum structure.

[0032] The weight of the leg-wheel robot 100 is mainly concentrated on the base 110 and the wheels 121, and the weight elements of the base 110 mainly include four motors that drive the thigh structure 122, a microcomputer, a circuit board, a motor, a battery, etc.

[0033] The dynamics model of the leg-wheel robot 100 is expressed by the following equation 1.

[0034]

number

[0035] m is the mass of the leg-wheel robot body, i.e., the base part, M is the mass of the wheels of the leg-wheel robot 100, a is the current height of the leg-wheel robot 100, i.e., the height of the leg-wheel robot 100 in the current thigh structure, and g is the gravitational acceleration. x is the distance that the midpoint of the wheel moves horizontally on the ground. If the wheel does not slip and does not leave the ground, x is equal to the linear distance traveled by the contact point of the wheel edge with the ground when the wheel motor rotates. (outside 1) JPEG0007639260000002.jpg5170 denotes the linear rotational speed of the wheel, (outside 2) JPEG0007639260000003.jpg8170 is the rotational linear acceleration; θ is the inclination angle of the leg-wheel robot 100; (Outside 3) JPEG0007639260000004.jpg9170 indicates the tilt angular velocity of the leg-wheel robot 100, (outside 4) JPEG0007639260000005.jpg9170 indicates the tilt angular acceleration of the leg-wheel robot 100. An example will be described in which the tilt angle θ is the pitch angle of the leg-wheel robot 100. u indicates the thrust applied to the leg-wheel robot 100, and there is a correspondence between u and the moment applied to the wheel.

[0036] Based on the above description, Fig. 3 is a flowchart of a control method for a leg-wheeled robot provided by an exemplary embodiment of the present application, the method being applied to and executed by a processor of the leg-wheeled robot. As shown in Fig. 3, the method includes the following steps: Step 320: Obtain the current motion state data of the leg-wheel type robot.

[0037] In some embodiments, basic data and current motion state data of the leg-wheel robot are obtained, and the basic data represents structural characteristics of the leg-wheel robot.

[0038] In general, the structural features of a leg-wheeled robot include physical values ​​of multiple parts of the robot. Taking the leg-wheeled biped robot of FIG. 1 as an example, the leg-wheeled robot includes, but is not limited to, a base part and a leg-wheel part, and the leg-wheel part and the base part have an inverted pendulum structure. That is, the control of the leg-wheel part links the control of the base part. The inverted pendulum structure is a linkage structure that controls the thigh structure and the base part by controlling the wheel. The structural features include the height, the base part mass, the wheel mass, the wheel radius, the distance between the two wheels, etc., indicated by the current thigh structure. The structural features also include the linkage structure between each part. For example, taking the leg-wheeled biped robot of FIG. 1 as an example, when the wheel is controlled to rotate clockwise and move to the right, the base part and the leg-wheel part move to the left relative to the wheel by linking them, and the linkage structure may be expressed in the basic data. This application does not limit the specific form or content of the basic data.

[0039] In some embodiments, the basic structure of the leg-wheel robot is fixed, so some basic data is set in advance. Taking the leg-wheel biped robot of FIG. 1 as an example, for example, the first mass data of the base part, the second mass data of the wheels in the leg-wheel part, the correspondence between the angle of the thigh structure and the height, etc. are shown. In some embodiments, some basic data is obtained by calculation. Taking the leg-wheel biped robot of FIG. 1 as an example, the height in the basic data is obtained by calculation according to the correspondence based on the angle of the shin segment and the thigh segment in the thigh structure.

[0040] In general, the current motion state data of the leg-wheel robot indicates the motion characteristics of the leg-wheel robot, and includes the joint angle and joint angular velocity of each joint of the leg-wheel robot, or may include other motion information of each joint according to actual needs, such as tilt angle, tilt angular velocity, wheel rotation distance (i.e., the travel distance of the leg-wheel robot), wheel rotation linear velocity, etc. This application does not limit the specific form or content of the current motion state data.

[0041] Exemplarily, the current motion state data includes base portion motion state data and leg wheel portion motion state data.

[0042] The base unit motion state data is information that represents the motion state of the base unit of the robot. For example, the base unit motion state data may include only the three-dimensional posture information of the robot, or posture information of the two-dimensional posture to be controlled in the three-dimensional posture information, and may further include the current height information of the base unit, thereby realizing the control of the height of the base unit.

[0043] The leg-wheel part motion state data is information representing a leg-wheel part motion state and a self-balancing state of the leg-wheel type robot, and may include, for example, at least a part of a current horizontal distance between the center of gravity of the leg-wheel part and the center of gravity of the base part, a current relative speed between the center of gravity of the leg-wheel part and the center of gravity of the base part, a motion position of the leg-wheel part, and a motion speed of the leg-wheel part, or may include other balance motion parameter information according to actual needs.

[0044] This application does not limit the specific configuration of the base portion motion state data and the leg wheel portion motion state data.

[0045] Exemplarily, step 320 is realized as follows: a sensor of the leg-wheel robot acquires sensor data of the leg-wheel robot, and then a processor of the leg-wheel robot estimates the motion state of the leg-wheel robot based on the sensor data to acquire current motion state data, where the sensor data includes at least one of the left wheel angle and left wheel angular velocity of the leg-wheel robot, at least one of the right wheel angle and right wheel angular velocity of the leg-wheel robot, the pitch angle and angular velocity of the base part, the roll angle and angular velocity of the base part, the yaw angle and angular velocity of the base part, and at least one of the angle and angular velocity of the joint motor of the leg-wheel robot.

[0046] Step 340: Input the current motion state data into the nonlinear controller to obtain the target joint angular acceleration reference value of the target robot joint of the leg-wheel robot.

[0047] Generally, the target robot joints of a leg-wheel robot are some or all of the joints of the leg-wheel robot.

[0048] For example, the current motion state data is input to the nonlinear controller, and the processor of the leg-wheel robot obtains the target joint angular acceleration reference value of the leg-wheel part of the leg-wheel robot; or the current motion state data is input to the nonlinear controller, and the processor obtains the target joint angular acceleration reference values ​​of all the joints of the leg-wheel robot, etc. The present application does not limit the number and range of the target robot joints.

[0049] Optionally, the target joint angular acceleration reference value of the target robot joint includes a leg-wheel section joint angular acceleration reference value of a leg-wheel type robot, or a base section joint angular acceleration reference value of a leg-wheel type robot and a leg-wheel section joint angular acceleration reference value of a leg-wheel type robot.

[0050] Optionally, when the leg-wheel type robot is a leg-wheel type biped robot, the leg-wheel section joint angular acceleration reference value of the leg-wheel type robot includes a left leg wheel section joint angular acceleration reference value and a right leg wheel section joint angular acceleration reference value.

[0051] As can be seen by referring to the above equation 1, the dynamics model of the leg-wheel robot itself is nonlinear, so first, a transformation process is performed on the above equation 1, and it is decomposed into the following equations 2 and 3.

[0052]

number

[0053]

number

[0054] v indicates the target joint angular acceleration reference value of the target robot joint of the leg-wheel robot, m refers to the mass of the leg-wheel robot body, i.e., the base part, M refers to the mass of the wheel of the leg-wheel robot, r is the radius of the wheel, l is the current height of the leg-wheel robot, i.e., the height of the leg-wheel robot in the current thigh structure, and g is the gravitational acceleration. (outside 5) JPEG0007639260000008.jpg9170 is the linear acceleration of the wheels. θ is the inclination angle of the leg-wheel robot. (outside 6) JPEG0007639260000009.jpg10170 The tilt angular velocity of the leg-wheel type robot is shown. (outside 7) JPEG0007639260000010.jpg10170 indicates the tilt angle acceleration of the leg-wheel robot. We will take the tilt angle θ as an example, which is the pitch angle of the leg-wheel robot. u indicates the balance moment, i.e., the thrust applied to the leg-wheel robot, and there is a correspondence between u and the moment applied to the wheel.

[0055] In some embodiments, based on the current motion state data, the processor of the leg-wheeled robot determines a target joint angular acceleration reference value of a target robot joint of the leg-wheeled robot through a nonlinear controller, with the target being that the linearized part of the dynamics relationship of the leg-wheeled robot matches the target balance state.

[0056] Exemplarily, the nonlinear controller is obtained by the steps of: determining an original energy relationship for the leg-wheeled robot, where the original energy relationship includes an original potential energy, and the original potential energy has a non-stationary characteristic; determining a variable energy relationship for the leg-wheeled robot, where the variable energy relationship includes a variable potential energy, and the variable potential energy has a stationary characteristic, and when the leg-wheeled robot is in a target balance state, the variable potential energy value is minimum; and determining a nonlinear controller with the original energy relationship change as a target being the variable energy relationship.

[0057] The nonlinear controller corresponds to the original energy relation and the variable energy relation.

[0058] Exemplarily, an original energy relationship and a variable energy relationship are determined. The original energy relationship includes original potential energy, and the original potential energy has a non-steady characteristic. The non-steady characteristic is that, in the calculation curve of the original potential energy, when the original potential energy is at a maximum value, the leg-wheel type robot is in a target balance state, and the maximum value situation of the original potential energy is unstable. The variable energy relationship includes variable potential energy, and the variable potential energy has a steady characteristic, and when the leg-wheel type robot is in the target balance state, the variable potential energy value is minimum. The steady characteristic is that, in the calculation curve of the variable potential energy, when the variable potential energy is at a minimum value, the leg-wheel type robot is in a target balance state, and the minimum value situation of the variable potential energy is stable.

[0059] Schematically, the original energy relationship is shown in Equation 4 below.

[0060]

number

[0061] q denotes momentum, p denotes angular momentum, (outside 8) JPEG0007639260000012.jpg11170, (outer 9) JPEG0007639260000013.jpg10170 and the inertia matrix (Outside 10) JPEG0007639260000014.jpg13170 and the original potential energy (Outside 11) JPEG0007639260000015.jpg10170 It is.

[0062] Generally, the variable energy relationship is shown in Equation 5 below.

[0063]

number

[0064] V d is variable potential energy, which has steady-state characteristics. For the expression of variable potential energy, refer to Equation 6.

[0065]

number

[0066] k is a free parameter, and when the variable potential energy is minimized, as can be proved by the Lyapunov function, the leg-wheeled robot is in the target balance state (Outside 12) JPEG0007639260000018.jpg11170 It can be stabilized.

[0067] M d is a preset matrix, and the pitch angle range (Outside 13) JPEG0007639260000019.jpg15170 is positive definite within the range, see Equation 7 below.

[0068]

number

[0069] m 11 teeth (Outside 14) JPEG0007639260000021.jpg14170 is a mathematical variable that represents m 12 teeth (Outside 15) JPEG0007639260000022.jpg14170 is a mathematical variable that represents m 22 teeth (Outside 16) JPEG0007639260000023.jpg12170 It is a mathematical variable that represents (Outside 17) JPEG0007639260000024.jpg10170 is a predefined free parameter, and k>0, (Outside 18) JPEG0007639260000025.jpg10170.

[0070] For example, in the embodiment of the present application, the original energy relationship change of Equation 4 is targeted to be the variable energy relationship of Equation 5, and the target joint angular acceleration reference value v of the target robot joint of the leg-wheel robot is determined.

[0071] When determining the target joint angular acceleration reference value of the target robot joint of the leg-wheeled robot, for different control targets, the processor of the leg-wheeled robot will determine the target joint angular acceleration reference value v of the leg-wheeled robot in different ways, which includes at least the following two situations: 1. Balance the leg-wheel robot on the balance point x * The control is performed so that the input voltage is fixed at .

[0072] In some embodiments, a target joint angular acceleration reference value of a target robot joint of a leg-wheel robot is determined according to Equation 8 below.

[0073]

number

[0074] (Outside 19) JPEG0007639260000027.jpg10170 and k v is a free parameter in the damping injection process, v denotes the target joint angular acceleration reference value of the target robot joint of the leg-wheel robot, x denotes the rotation distance of the leg-wheel robot wheel, and x * represents the balance point of the wheel in the target balance state, θ represents the tilt angle of the leg-wheel robot, and A1, A2, A3, and A4 are pre-established parameter calculation expressions. In this way, when controlling a leg-wheel robot, the leg-wheel robot is balanced by the balance moment to reach the balance point x * The control is performed so that the input voltage is fixed at .

[0075] Regarding the calculation methods of the above A1, A2, A3, and A4, the following formulas 9 to 12 may be referred to.

[0076]

number

[0077]

number

[0078]

number

[0079]

number

[0080] m 11 , m 12 , m 22 For this, we refer to the above formula 7, (outside 20) JPEG0007639260000032.jpg17170 indicates the derivative of the logarithmic function F(θ) with respect to θ, and the calculation method of F(θ) is shown in 13 below.

[0081]

number

[0082] α 11 and α 12 Regarding the calculation method, please refer to the following formula 14.

[0083]

number

[0084] α 11 is the first row obtained by the above matrix multiplication, and α 12 is the second row obtained by the above matrix multiplication. Based on the above equation 8, the current distance from the wheel to the balance point x * Calculate the target joint angular acceleration reference value to adjust the target robot joint of the leg-wheel robot corresponding to the balance point [0 0x * 0] T That is, the H in Equation 5 is controlled to be fixed according to the current target joint angular acceleration reference value obtained by the calculation. d but (outside 21) JPEG0007639260000035.jpg11170 and (outside 22) JPEG0007639260000036.jpg11170 Fulfilling (outside 23) JPEG0007639260000037.jpg10170 indicates that the tilt angle is 0, (outside 24) JPEG0007639260000038.jpg9170 The leg-wheel type robot has a balance point x * The balance point x * is a value specified in real time, for example, a value determined based on a control operation for a leg-wheel type robot.

[0085] 2. Control the leg-wheeled robot to move at a constant speed in a target balance state. In some embodiments, the target joint angular acceleration reference is determined based on Equation 15 below.

[0086]

number

[0087] (Outside 25) JPEG0007639260000040.jpg8170 θ is the inclination angle of the leg-wheel type robot, and k v are free parameters in the damping injection process, and A1, A2, A3, and A4 are pre-established parameter calculation expressions, and Equations 9 to 12 may be referred to. In this way, when controlling a leg-wheeled robot, the leg-wheeled robot is controlled in a target balance state by the balance moment. (outside 26) JPEG0007639260000041.jpg9170 The object is controlled to move at a uniform speed.

[0088] Based on the above equation 15, the current wheel linear velocity (outside 27) JPEG0007639260000042.jpg9170 from target speed (outside 28) JPEG0007639260000043.jpg8170 Calculate the target joint angular acceleration reference value v of the leg-wheel type robot to adjust the target robot joint of the leg-wheel type robot corresponding to (outside 29) JPEG0007639260000044.jpg9170 In other words, the currently calculated target joint angular acceleration reference value v is used to control the H d but (Outside 30) JPEG0007639260000045.jpg11170 and (Outside 31) JPEG0007639260000046.jpg10170 Fulfilling (Outside 32) JPEG0007639260000047.jpg10170 indicates that the tilt angle is 0, (Outside 33) JPEG0007639260000048.jpg9170 is the motion speed of the leg-wheel type robot (Outside 34) JPEG0007639260000049.jpg10170 This indicates that the balance point (Outside 35) JPEG0007639260000050.jpg8170 is a value specified in real time, for example, a value determined based on a control operation for a leg-wheel type robot.

[0089] Step 360: Input the target joint angular acceleration reference value into a whole-body dynamics controller to output joint moments for controlling the leg-wheeled robot to perform the control task.

[0090] The whole-body dynamics controller includes a system of equations consisting of a set of control tasks and constraints.

[0091] Assuming that the constraint conditions are satisfied, the processor of the leg-wheeled robot solves a quadratic optimization problem using the target joint angular acceleration reference value as the control target through the whole-body dynamics controller to obtain the joint moments for controlling the leg-wheeled robot to perform the control task.

[0092] In general, the constraints include at least two of a friction cone constraint, a closed loop link constraint, and a dynamics constraint.

[0093] The friction cone constraint is used to constrain the load-bearing surface reaction forces to satisfy the friction cone, where the load-bearing surface reaction forces are the acting forces of the load-bearing surface on the leg-wheeled robot, the closed-loop link constraint is used to constrain the thigh link structure of the leg-wheeled robot, and the dynamics constraint is used to constrain the control task based on the dynamics model of the leg-wheeled robot.

[0094] The control tasks include self-balancing tasks and motion tasks, etc. The motion tasks can be divided into static motion tasks and dynamic motion tasks.

[0095] The static motion tasks include a wheel roll task of the leg-wheel robot, a base height task of the leg-wheel robot, a base roll task, a base pitch task, a base yaw task, a base tail control task, and a leg-wheel balance task.

[0096] Dynamic movement tasks include jumping tasks, dual-wheel alternating walking tasks, and obstacle avoidance tasks where one wheel is supported while the other wheel is off the ground.

[0097] As described above, according to the control method for a leg-wheeled robot provided in the embodiment of the present application, the current motion state data of the leg-wheeled robot is obtained and input to the nonlinear controller, the processor of the leg-wheeled robot obtains a target joint angular acceleration reference value of a target robot joint of the leg-wheeled robot, and the target joint angular acceleration reference value of the target robot joint of the leg-wheeled robot determined by the nonlinear controller is input to the whole-body dynamics controller, and the processor can obtain a joint moment for controlling the leg-wheeled robot to perform a control task. The method provided in the embodiment of the present application combines a nonlinear control method and a whole-body dynamics control method, and in the control process for the robot, not only can the nonlinear control method improve the accuracy of balance control for the leg-wheeled robot, but also the whole-body dynamics method can realize compliance control of the robot, thereby improving the flexibility and control effect of compliance control for the robot.

[0098] Preferably, the embodiments of the present application further provide a method for determining a nonlinear controller.

[0099] Preferably, in the embodiment of the present application, two determination methods of the target joint angular acceleration of the target robot joint are respectively proposed for different control targets, and the leg-wheeled robot is balanced to the balance point x * When the control target is to control the leg-wheeled robot to move at a constant speed in the target balance state, the target joint angular acceleration reference value of the target robot joint of the leg-wheeled robot can be determined by Equation 8. When the control target is to control the leg-wheeled robot to move at a constant speed in the target balance state, the target joint angular acceleration reference value is determined by Equation 15.

[0100] Preferably, the embodiments of the present application further propose a method for acquiring current motion state data of the leg-wheeled robot, estimating the motion state of the leg-wheeled robot based on sensor data, acquiring the current motion state data, and improving the accuracy of data determination.

[0101] The above embodiment shows the process of obtaining a target joint angular acceleration reference value of a target robot joint of a leg-wheeled robot by a nonlinear controller based on current motion state data, inputting the target joint angular acceleration reference value into a whole-body dynamics controller, and then obtaining the joint moment of each joint when the leg-wheeled robot performs a control task. The following embodiment takes the control task as a self-balancing task as an example to describe in detail the method for determining the joint moment of each joint of a leg-wheeled robot by a whole-body dynamics controller.

[0102] 4 is a flowchart of a control method for a leg-wheeled robot provided by an exemplary embodiment of the present application, the method being applied to and executed by a processor of the leg-wheeled robot. As shown in FIG. 4, the method includes the following steps: Step 420: Obtain current motion state data of the leg-wheel robot; The method for acquiring the motion state data has already been explained in step 320 above, so it will not be described in detail here. Step 440: input the current motion state data into the nonlinear controller to obtain the target joint angular acceleration reference value of the target robot joint of the leg-wheel robot; The method for determining the target joint angular acceleration reference value based on the current motion state data has already been described in step 440 above, so no further description will be given here. Step 460: Input the target joint angular acceleration reference values ​​into a whole-body dynamics controller to output joint moments for controlling the leg-wheeled robot to perform a self-balancing task.

[0103] Exemplarily, taking the control task as an example where the control task is a self-balancing task, the controller of the leg-wheeled robot inputs the target joint angular acceleration reference value of the target robot joint of the leg-wheeled robot into the whole-body dynamics controller, and searches for an optimal solution that is closest to the target joint angular acceleration reference value for the leg-wheeled robot to perform the self-balancing task as the output of the whole-body dynamics controller, and the output joint moment controls the leg-wheeled robot to perform the self-balancing task.

[0104] As an example, taking the control task as a self-balancing task, the following sub-steps 4601 to 4602 show the process of obtaining the joint moment by performing calculations using the target joint angular acceleration reference value of the target robot joint of the leg-wheel robot input to the whole-body dynamics controller. Step 4601: Based on the target joint angular acceleration reference values ​​of the leg-wheel parts, a target posture angular acceleration reference value of the base part is calculated with the target being to maintain relative balance.

[0105] The target joint angular acceleration reference value of the leg wheel unit determined in step 440 is input to the whole body dynamics controller. Since the base unit is connected to the leg wheel unit by a holder, or the connection between the base unit and the leg wheel unit is realized in other ways, the target attitude angular acceleration reference value of the base unit is determined according to the target joint angular acceleration reference value of the leg wheel unit.

[0106] In order to enable the leg-wheel type robot to achieve a self-balancing task, during the robot's motion process, the horizontal distance between the center of gravity of the leg-wheel part and the center of gravity of the base part is controlled to be smaller than a target horizontal distance threshold, and the relative speed between the center of gravity of the leg-wheel part and the center of gravity of the base part is controlled to be smaller than a target relative speed threshold, where the target horizontal distance threshold indicates the horizontal distance between the center of gravity of the leg-wheel part and the center of gravity of the base part that keeps the center of gravity of the entire leg-wheel type robot stable, and the target relative speed threshold indicates the relative speed between the center of gravity of the leg-wheel part and the center of gravity of the base part that keeps the leg-wheel type robot stable.

[0107] For example, in order to maintain relative balance in a leg-wheel type robot, during the robot's movement process, the horizontal distance between the center of gravity of the leg-wheel part of the leg-wheel type robot and the center of gravity of the base part is controlled to be smaller than the minimum value of the distance between the center of gravity of the leg-wheel part and each edge of the leg-wheel part, and the relative speed between the center of gravity of the leg-wheel part of the leg-wheel type robot and the center of gravity of the base part is controlled to be 0 or close to 0, so that the center of gravity of the base part is always located within the area of ​​the leg-wheel part and there is no relative motion between the base part and the leg-wheel part, thereby ensuring that the leg-wheel type robot maintains a balanced state during the movement process.

[0108] The target attitude angular acceleration reference value of the base part is obtained by calculation based on the target joint angular acceleration reference value of the leg-wheel part, the horizontal distance between the center of gravity of the leg-wheel part and the center of gravity of the base part, and the relative speed between the center of gravity of the leg-wheel part and the center of gravity of the base part. The center of gravity of the leg-wheel part is set to maintain the relative balance of the leg-wheel type robot.

[0109] Step 4602: Assuming that the constraints are satisfied, the whole-body dynamics controller determines the joint moments of each joint when the leg-wheel robot performs a self-balancing task based on the target joint angular acceleration reference value of the leg-wheel section and the target attitude angular acceleration reference value of the base section.

[0110] For example, a target motion acceleration of the leg-wheel type robot is determined based on a target joint angular acceleration reference value of the leg-wheel section and a target attitude angular acceleration reference value of the base section, a target motion acceleration estimate of the leg-wheel type robot is determined based on current motion state data, and, assuming that the constraint conditions are satisfied, joint moments of each joint when the leg-wheel type robot performs a self-balancing task are determined based on the target motion acceleration and target motion acceleration estimate of the leg-wheel type robot.

[0111] The joint moments are used as the output of a whole-body dynamics controller to control each joint motor of a leg-wheeled robot, allowing the robot to perform self-balancing tasks.

[0112] For example, the processor of the leg-wheel type robot determines the joint moment of each joint based on the joint target motion acceleration of each joint of the leg-wheel type robot, performs friction compensation on the joint moment, generates a joint control current value based on the joint moment after the compensation friction, and sends the joint control current value to a corresponding joint control motor to output a corresponding moment from the motor, thereby realizing control of each joint of the leg-wheel type robot, and further realizing motion control of the leg-wheel type robot.

[0113] As described above, in the control method for a leg-wheeled robot provided by this embodiment, when the control task includes a self-balancing task, the target joint angular acceleration reference value of the target robot joint of the leg-wheeled robot determined by the nonlinear controller is used as the input of the whole-body dynamics controller, and the processor of the leg-wheeled robot determines the target motion acceleration when the leg-wheeled robot performs the self-balancing task, and converts the target motion acceleration of the leg-wheeled robot into the joint target motion acceleration of each joint of the leg-wheeled robot according to the constraint conditions, and finally converts it into the joint moment of each joint for output. The embodiment of the present application provides a method for calculating the joint moment of each joint when the leg-wheeled robot performs the self-balancing task, and mainly uses the whole-body dynamics controller to perform analysis processing on the nonlinear joint angular acceleration to obtain the joint moment of each joint for controlling the leg-wheeled robot to perform the task, thereby realizing the optimization control of the compliance control of the leg-wheeled robot in accurately performing the balance control on the leg-wheeled robot.

[0114] The above embodiment shows the process of the leg-wheeled robot control method combining the nonlinear control method and the whole body dynamics method, which combines the advantages of the nonlinear control and the whole body dynamics control method to provide a more optimized control method. The following embodiment shows the implementation process of obtaining the motion reference trajectory by adding user control input or based on the motion state input based on the embodiment shown in Figures 3 and 4.

[0115] 5 is a flowchart of a control method for a leg-wheeled robot provided by an exemplary embodiment of the present application, the method being applied to and executed by a processor of the leg-wheeled robot. As shown in FIG. 5, the method includes the following steps: Step 520: Obtain current motion state data of the leg-wheel robot; The method for acquiring the motion state data has already been explained in step 320 above, so it will not be described in detail here. Step 542a: input the current motion state data into the nonlinear controller to obtain the target joint angular acceleration reference value of the target robot joint of the leg-wheel robot; The method for determining the target joint angular acceleration reference value of the target robot joint of the leg-wheel robot based on the current motion state data has already been described in step 440 above, so no further description will be given here. Step 542b: Obtain a movement reference trajectory for the movement task.

[0116] In general, the motion reference trajectory of a motion task represents desired motion information of the leg-wheel robot at a target timing, which may be the timing next to the current timing or a timing after a preset time interval from the current timing.

[0117] Exemplarily, the motion reference trajectory includes information on a desired motion state and a desired self-balancing state of the leg-wheel robot at a target timing, such as at least a part of a target horizontal distance between the center of gravity of the leg wheel part and the center of gravity of the base part, a target relative speed between the center of gravity of the leg wheel part and the center of gravity of the base part, a target motion position of the leg wheel part, and a target motion speed of the leg wheel part, or may include other motion parameter information according to actual conditions.

[0118] The processor of the leg-wheel robot obtains the motion reference trajectory of the motion task based on the user control input, or inputs the current motion state data into the trajectory generator of the motion task to obtain the motion reference trajectory of the motion task.

[0119] Exemplarily, the processor receives a user control input from a remote control and determines a motion reference trajectory for a motion task based on the user control input, or inputs the current motion state data into a walking motion trajectory generator to obtain a motion reference trajectory for controlling the leg-wheeled robot to perform a dual-wheel alternating walking task, or inputs the current motion state data into a jumping trajectory generator to obtain a motion reference trajectory for controlling the leg-wheeled robot to perform a jumping task.

[0120] In general, the movement reference trajectory of the dual-wheel alternating walking task includes at least one of a movement reference trajectory when the left thigh supports and the right thigh rises, a movement reference trajectory when the right thigh supports and the left thigh lands, a movement reference trajectory when switching from left thigh support to right thigh support, and a movement reference trajectory when switching from right thigh support to left thigh support.

[0121] In general, the motion reference trajectory of the jump task includes at least one of a motion reference trajectory of a takeoff staircase, a motion reference trajectory of a midair staircase, and a motion reference trajectory of a landing staircase.

[0122] In general, taking the current motion state data as input to a jump trajectory generator to obtain a motion reference trajectory for controlling a leg-wheeled robot to perform a jump task as an example, the process of obtaining a motion reference trajectory for a motion task includes the following steps: First step: Obtain a simplified model of the leg-wheel robot performing a jump task. For example, simplifying the thigh of a leg-wheel robot into an extendable and powerable rod, simplifying the base of a leg-wheel robot into a floating base, simplifying the tail and base of a leg-wheel robot into a single rigid body, etc.; Second step: The current motion state data is substituted into the dynamics model of the leg-wheeled robot performing the jump task to obtain the dynamics equations for performing the jump task. In general, the current motion state data includes joint angles and angular velocities of the leg-wheel robot. Third step: Set up constraints. In general, the constraints include friction cone constraints of different stairs, closed-loop link constraints, dynamics constraints, path constraints, boundary constraints, and continuity constraints between different stairs, etc. When the wheels of the leg-wheeled robot are on the ground, the constraints include friction cone constraints, closed-loop link constraints and dynamics constraints, and when the wheels of the leg-wheeled robot are not on the ground, the constraints include closed-loop link constraints and dynamics constraints. Fourth step: Obtain the motion reference trajectory for the leg-wheel robot to perform the jump task by calculation.

[0123] An optimal solution for the leg-wheel robot to execute a jump task is selected based on the constraint conditions, and planned position and posture data at multiple time points during the execution of the jump task by the leg-wheel robot, i.e., a motion reference trajectory for the leg-wheel robot to execute the jump task, is determined. Exemplarily, the multiple time points are multiple time points with a control step size of 1 ms, and each time point corresponds to a set of planned position data of the leg-wheel robot.

[0124] This application does not limit the specific manner of acquiring the motion reference trajectory of the motion task.

[0125] Step 560: The target joint angular acceleration reference values ​​and the motion reference trajectory are input to a whole body dynamics controller to output joint moments for controlling the leg-wheeled robot to perform the motion task.

[0126] In general, the motion tasks performed by the leg-wheel robot include at least one of a wheel roll task of the leg-wheel robot, a base height task of the leg-wheel robot, a base roll task, a base pitch task, a base yaw task, a base tail control task, and a leg-wheel balance task.

[0127] In general, the constraints include at least two of a friction cone constraint, a closed-loop link constraint, and a dynamics constraint, where the friction cone constraint is used to constrain the load-bearing surface reaction force to satisfy the friction cone, where the load-bearing surface reaction force is the acting force of the load-bearing surface on the leg-wheeled robot, the closed-loop link constraint is used to constrain the thigh link structure of the leg-wheeled robot, and the dynamics constraint is used to constrain the control task based on the dynamics model of the leg-wheeled robot.

[0128] When the wheels of the leg-wheeled robot are on the ground, the constraints include friction cone constraints, closed-loop link constraints and dynamics constraints, and when the wheels of the leg-wheeled robot are not on the ground, the constraints include closed-loop link constraints and dynamics constraints.

[0129] Assuming that the constraint conditions are satisfied, the processor of the leg-wheeled robot uses the whole-body dynamics controller to solve a quadratic optimization problem using the joint angular accelerations and the motion reference trajectory as control targets to obtain joint moments for controlling the leg-wheeled robot to perform the motion task.

[0130] For example, the motion reference trajectory is a motion reference trajectory for controlling a leg-wheeled robot to perform a dual-wheel alternating walking task obtained by the walking motion trajectory generator. The whole body dynamics controller calculates the target joint angular acceleration reference value of the target robot joint of the leg-wheeled robot and the motion reference trajectory input, and the process of obtaining the joint moment can be performed by referring to the following steps 5601 to 5602: Step 5601: Based on the target joint angular acceleration reference values ​​of the leg-wheel parts, a target posture angular acceleration reference value of the base part is calculated with the motion reference trajectory as a target.

[0131] The target joint angular acceleration reference values ​​of the leg-wheel sections determined in step 542a are input to the whole body dynamics controller.

[0132] The base is connected to the leg-wheel unit by a holder or by other means to realize the connection between the base and the leg-wheel unit, and in order to make the leg-wheel robot realize the motion task, the target joint angular acceleration of the leg-wheel robot is controlled during the robot's motion process to realize the posture adjustment and motion task of the base. Therefore, the processor of the leg-wheel robot determines the target posture angular acceleration reference value of the base according to the target joint angular acceleration reference value of the leg-wheel unit and with the motion reference trajectory as the target.

[0133] The motion reference trajectory is a sequence including planned position and posture data for the leg-wheeled robot at multiple points in time, where the planned position and posture data includes data such as the target posture of the base part of the leg-wheeled robot.

[0134] Illustratively, the processor of the leg-wheel type robot determines a base part target posture of the leg-wheel type robot at a target timing from a posture data sequence indicated by the motion reference trajectory, and combines the target joint angular acceleration reference values ​​of the leg-wheel part to determine a target posture angular acceleration reference value of the base part.

[0135] Step 5602: Assuming that the constraints are satisfied, the whole-body dynamics controller determines the joint moments of each joint when the leg-wheel robot performs the movement task based on the target joint angular acceleration reference value of the leg-wheel section and the target attitude angular acceleration reference value of the base section.

[0136] For example, the processor of the leg-wheel type robot determines a target motion acceleration of the leg-wheel type robot based on the target joint angular acceleration reference value of the leg-wheel section and the target attitude angular acceleration reference value of the base section, and based on the current motion state data, the processor determines a target motion acceleration estimate of the leg-wheel type robot, and based on the target motion acceleration and the target motion acceleration estimate of the leg-wheel type robot, assuming that the constraint conditions are satisfied, the processor determines the joint moment of each joint when the leg-wheel type robot performs the motion task.

[0137] The joint moments are used as the output of a whole-body dynamics controller to control each joint motor of a leg-wheeled robot, enabling the robot to perform a dual-wheel alternating walking task.

[0138] For example, the processor of the leg-wheel type robot determines the joint moment of each joint based on the joint target motion acceleration of each joint of the leg-wheel type robot, performs friction compensation on the joint moment, generates a joint control current value based on the joint moment after the compensation friction, and sends the joint control current value to a corresponding joint control motor to output a corresponding moment via the motor, thereby realizing control of each joint of the leg-wheel type robot, and further realizing motion control of the leg-wheel type robot.

[0139] As described above, in the control method for a leg-wheeled robot provided by the embodiment of the present application, when the control task includes a motion task, the processor of the leg-wheeled robot inputs the target joint angular acceleration reference value of the target robot joint of the leg-wheeled robot and the motion reference trajectory output from the nonlinear controller into the whole-body dynamics controller to obtain the joint moment for controlling the leg-wheeled robot to execute the control task. The method provided by the embodiment of the present application not only realizes control of the execution of motion tasks by the leg-wheeled robot after completing the basic target self-balancing task of the leg-wheeled robot, but also combines the nonlinear control method with the whole-body dynamics control method to enable the leg-wheeled robot to obediently, accurately and stably complete each motion task, thereby improving the control accuracy of the leg-wheeled robot.

[0140] Preferably, the embodiment of the present application further provides two different acquisition ways of the motion reference trajectory of the motion task to ensure the accuracy of the motion reference trajectory: the motion reference trajectory can be determined by receiving a user control input from a remote control, or the motion reference trajectory can be obtained by inputting the current motion state data into a trajectory generator of the motion task, and the trajectory generator includes a walking motion trajectory generator and / or a jumping trajectory generator.

[0141] Preferably, the embodiments of the present application further provide specific constraints included in the constraint conditions, and specific constraint conditions when the wheels of the leg-wheel robot are on the ground and not on the ground, to provide an accurate basis for determining the target joint angular acceleration reference value of the target robot joint.

[0142] The only difference between sub-step 4602 in the embodiment of Fig. 4 and sub-step 5602 in the embodiment of Fig. 5 is the different control tasks realized by the leg-wheel robot, but the realization processes of the steps in the two embodiments are similar. Fig. 6 is a flowchart of a method for determining joint moments of each joint when the leg-wheel robot performs a control task by a leg-wheel robot provided in one exemplary embodiment of the present application, which is applied to and executed by a processor of the leg-wheel robot.

[0143] Below, using the embodiment of Figure 6, we will explain in detail the process of determining the joint moment of each joint when a leg-wheel robot performs a control task by the whole-body dynamics controller based on the target joint angular acceleration reference value of the leg-wheel section and the target attitude angular acceleration reference value of the base section, assuming that the constraint conditions are satisfied.

[0144] Step 6602-1: Determine a target motion acceleration of the leg-wheel robot based on the target joint angular acceleration reference value of the leg-wheel section and the target attitude angular acceleration reference value of the base section.

[0145] In general, the target movement acceleration of a leg-wheel type robot is the amount of acceleration possessed by the entire leg-wheel type robot obtained after comprehensively taking into consideration the balance movement process of the leg-wheel type robot and the posture adjustment process of the leg-wheel type robot base part.

[0146] For example, a normalization process is performed on the base target attitude angular acceleration reference value and the leg-wheel target joint angular acceleration reference value of the leg-wheel robot based on a predetermined weight, and the acceleration vector or acceleration matrix obtained after the normalization process is set as the target motion acceleration of the leg-wheel robot. The present application does not limit the specific manner of generating the target motion acceleration of the leg-wheel robot, or the specific configuration and number of dimensions of the target motion acceleration of the leg-wheel robot.

[0147] Step 6602-2: Determine a target motion acceleration estimate for the leg-wheel robot based on the current motion state data.

[0148] According to the corresponding relationship between the joint motion process of the leg-wheel robot and the dynamic motion process of the leg-wheel robot, the processor of the leg-wheel robot can substitute the current motion state data into a preset equation to calculate and generate an estimated value of the target motion acceleration of the leg-wheel robot, or can generate the estimated value of the target motion acceleration of the leg-wheel robot in other ways. The present application does not limit the specific process of generating the estimated value of the target motion acceleration.

[0149] Here, the present application does not limit the order of execution of steps 6602-1 and 6602-2, and step 6602-1 may be executed before step 6602-2, or step 6602-2 may be executed before step 6602-1, or step 6602-1 and step 6602-2 may be executed simultaneously.

[0150] Step 6602-3: Assuming that the constraints are satisfied, determine the joint moments of each joint when the leg-wheeled robot performs the control task based on the target motion acceleration of the leg-wheeled robot and the estimated target motion acceleration.

[0151] In general, the constraint conditions are used to indicate the constraint relationship between the motion information and the joint motion information of the leg-wheel robot. Specifically, the constraint conditions are used to limit the constraint conditions and constraint relationships that each joint of the leg-wheel robot needs to satisfy in the motion process of the leg-wheel robot, so that the physical motion information of the leg-wheel robot and the joint motion information of the leg-wheel robot are mapped to each other and the data obtained after the mapping is within a reasonable range that satisfies the physical motion constraints.

[0152] For example, the constraint function may be formed based on the dynamics equations and constraint conditions of the leg-wheel robot, or may be generated by calculating based on the kinematics equations using a preset algorithm. For example, the constraint function may further include one or more of an equality constraint subfunction, an inequality constraint subfunction, and a threshold constraint subfunction. The present application does not limit the specific configuration or the specific generation method of the constraint function.

[0153] Exemplarily, determining the joint moment of each joint of a leg-wheel robot using a constraint function specifically includes the following steps: for each joint, first generate an error function based on the target motion acceleration of the leg-wheel robot and an estimated amount of the target motion acceleration of the leg-wheel robot; then determine the joint angular acceleration that satisfies the constraint function and minimizes the error function as the target joint angular acceleration based on the error function and the constraint function; and finally determine the joint moment based on the target joint angular acceleration.

[0154] The process is explained in detail below: (1) Generate an error function based on the target motion acceleration of the leg-wheel robot and an estimate of the target motion acceleration of the leg-wheel robot; The error function is used to indicate the difference between the target motion acceleration of the leg-wheel robot and the estimated amount of target motion acceleration of the leg-wheel robot, and the numerical value of the error function is used to indicate the magnitude of the difference between the target motion acceleration of the leg-wheel robot and the estimated amount of target motion acceleration of the leg-wheel robot.

[0155] For example, the difference between the target motion acceleration of the leg-wheel robot and the estimated amount of the target motion acceleration is the error function, and since the target motion acceleration of the leg-wheel robot is a function of the target joint angular acceleration reference value of the leg-wheel robot, the generated error function is also a function of the target joint angular acceleration reference value of the leg-wheel robot, or the target motion acceleration of the leg-wheel robot and the estimated amount of the target motion acceleration are substituted into a preset equation system or a preset algorithm to generate the error function. The present application does not limit the method of generating the error function.

[0156] (2) Based on the error function and the constraint function, a joint angular acceleration that satisfies the constraint function and minimizes the error function is determined as a target joint angular acceleration.

[0157] For example, the constraint function may include the joint angular acceleration parameter amount of the leg-wheel robot, or the predetermined function may include variables such as joint moment parameter amount associated with the joint angular acceleration variable. As described above, the error function is also a function of the target joint angular acceleration reference value, and in this case, by adjusting the joint angular acceleration variable of the robot and the other parameter amount associated with the joint angular acceleration variable, the error function has a minimum value and the value of the joint angular acceleration variable can meet the requirement of the constraint function. In this way, the target joint angular acceleration can be obtained, which not only meets the posture control requirements of the leg-wheel robot base part and the balance motion control requirements of the leg-wheel robot, but also meets the leg-wheel robot kinematic equation and its constraint conditions, thereby effectively avoiding the problem of invalid or lapsed control that occurs when only the leg-wheel robot motion acceleration is simply projected into the joint angular acceleration space, which is caused by the joint angular acceleration obviously exceeding the motor control range or the acquired leg-wheel robot joint configuration posture obviously not matching the motion state of the leg-wheel robot.

[0158] (3) determining a joint moment based on the target joint angular acceleration; The processor of the leg-wheel robot determines the joint moment of the joint by substituting the acquired target joint angular acceleration of the joint of the leg-wheel robot into a predetermined function, which is used to indicate the conversion relationship between the target joint angular acceleration and the joint moment.

[0159] As described above, according to the method for determining the joint moments of a leg-wheel type robot in this embodiment, the target motion acceleration and an estimated amount of the target motion acceleration of the leg-wheel type robot are determined, and assuming that the constraint conditions are satisfied, the processor of the leg-wheel type robot determines the joint moments of each joint of the leg-wheel type robot, thereby obtaining an optimal solution to the quadratic optimization problem, thereby providing a method for determining the joint moments of each joint of the leg-wheel type robot.

[0160] For example, the leg-wheel robot has a dual-wheel (left and right leg-wheel) configuration. The process of calculating the target motion acceleration of the leg-wheel of the leg-wheel robot will be described below.

[0161] First, in order to realize the control of the balance motion of the leg-wheel type robot, the relative acceleration between the base part (base) of the leg-wheel type robot and the leg-wheel part (wheel) of the leg-wheel type robot is calculated. (Outside 36) JPEG0007639260000051.jpg8170 By installing the above, the relative displacement between the base unit and the leg-wheel unit of the leg-wheel type robot is (Outside 37) JPEG0007639260000052.jpg9170 In this case, the non-minimum phase system formed by the leg-wheel section and the base section is at a balance point. Thus, the following feedback control rate can be established:

[0162]

number

[0163] (Outside 38) JPEG0007639260000054.jpg13170 is the relative acceleration between the base part of the leg-wheel type robot and the leg-wheel part of the leg-wheel type robot, (Outside 39) JPEG0007639260000055.jpg12170 is the target horizontal distance between the center of gravity of the leg-wheel part and the center of gravity of the base part of the leg-wheel type robot, (outside 40) JPEG0007639260000056.jpg10170 is the current horizontal distance between the center of gravity of the leg-wheel part of the leg-wheel type robot and the center of gravity of the base part, where "0" indicates that the target relative velocity between the center of gravity of the leg-wheel part of the leg-wheel type robot and the center of gravity of the base part is 0, (outside 41) JPEG0007639260000057.jpg9170 represents the current relative velocity between the center of gravity of the leg-wheel part and the center of gravity of the base part of the leg-wheel type robot, (outside 42) JPEG0007639260000058.jpg12170 , (outside 43) JPEG0007639260000059.jpg8170 are parameter amounts selected according to actual needs. Since the leg-wheel robot adopts a dual-wheel configuration, the "leg-wheel" here is a virtual single wheel obtained by fitting the left leg wheel part and the right leg wheel part, and the "leg-wheel part gravity center" is the gravity center of the virtual single wheel obtained by fitting the left leg wheel part and the right leg wheel part.

[0164] Based on the above Equation 16, the relative acceleration of the leg wheel part (virtual single wheel) with respect to the base part is calculated to realize the control task. (outside 44) JPEG0007639260000060.jpg10170 In the leg-wheel type robot, the leg-wheel unit (virtual single wheel) acceleration and the relative acceleration of the leg-wheel unit (virtual single wheel) with respect to the base unit satisfy the following relationship:

[0165]

number

[0166] (outside 45) JPEG0007639260000062.jpg10170 is the motion acceleration of the leg wheel part (virtual single wheel), (outside 46) JPEG0007639260000063.jpg11170 is the relative acceleration of the leg wheel part (virtual single wheel) with respect to the base part, (outside 47) JPEG0007639260000064.jpg12170 is the distance between the base part and the leg wheel part (virtual single wheel) along the x coordinate axis direction (hereinafter also referred to as the x direction), (outside 48) JPEG0007639260000065.jpg9170 is the distance between the base part and the leg-wheel part (virtual single wheel) along the z coordinate axis direction (hereinafter also referred to as the z direction). The positive direction of the x coordinate axis is the forward direction of the leg-wheel type robot, the positive direction of the z coordinate axis is the vertically upward direction, and g is the gravitational acceleration.

[0167] (outside 49) JPEG0007639260000066.jpg8170 In this case, an external annular controller can be designed to control the linear acceleration of the leg wheel part (virtual single wheel), and further control the x-directional motion of the leg wheel part (virtual single wheel), and the specific control rate is as follows:

[0168]

number

[0169] (outside 50) JPEG0007639260000068.jpg10170 is the distance in the x direction between the base part and the leg wheel part (virtual single wheel) at the current timing, and can be calculated based on, for example, the feedback rate, (outside 51) JPEG0007639260000069.jpg8170 is the distance in the z direction between the base part and the leg wheel part (virtual single wheel) at the current timing, g is the gravitational acceleration, (outside 52) JPEG0007639260000070.jpg9170 is the target position of the leg wheel part (virtual single wheel), (outside 53) JPEG0007639260000071.jpg9170 is the current position of the leg wheel part (virtual single wheel), (outside 54) JPEG0007639260000072.jpg10170 is the target speed of the leg wheel part (virtual single wheel), (outside 55) JPEG0007639260000073.jpg8170 is the current speed of the leg wheel (virtual single wheel), (outside 56) JPEG0007639260000074.jpg8170 , (outside 57) JPEG0007639260000075.jpg9170 are parameter amounts selected according to actual needs.

[0170] Therefore, we finally obtain the feedback ratio that balances the controller, and the feedback coefficient K p,1 , K p,2 , K d,1 , Kd,2 can be calculated by designing the poles based on the inner and outer annular systems. For example, the feedback coefficients are p,1 =1200,K d,1 =300, K p,2 ,=1,K d,2 = 0.2, which gives the following expression:

[0171]

number

[0172] (outside 58) JPEG0007639260000077.jpg12170 is the motion acceleration of the leg wheel part (virtual single wheel) at the target timing, (outside 59) JPEG0007639260000078.jpg11170 is a desired distance in the x direction between the base unit and the leg wheel unit (virtual single wheel) at the target timing, which can be calculated based on, for example, a feedback rate, (outside 60) JPEG0007639260000079.jpg10170 is the distance in the x direction between the base part and the leg wheel part (virtual single wheel) at the current timing. For the meanings of the other parameter amounts, see above.

[0173] Then, the x component in the base operation matrix is ​​cancelled, and the linear acceleration of the left and right leg wheel parts is substituted for it. The acceleration of the virtual single wheel equivalent to the left and right leg wheel parts affects the balance, and the differential speed affects the yaw angular acceleration of the leg wheel part (virtual single wheel). Specifically, the yaw angular acceleration satisfies the following dynamics equation.

[0174]

number

[0175] (outside 61) JPEG0007639260000081.jpg10170 is the target relative yaw angular acceleration, (outside 62) JPEG0007639260000082.jpg9170 is the target relative yaw angle, and 0 indicates that the desired relative yaw angular velocity at the target timing (i.e., the target relative yaw angular velocity) is 0. (outside 63) JPEG0007639260000083.jpg9170 is now the relative yaw angle, (outside 64) JPEG0007639260000084.jpg9170 is the current relative yaw angular velocity and K p,3 , K d,3 For example, K p,3 = 100, K d,3 =100.

[0176] In addition, based on the structural relationship of the dual-wheel type leg-wheel unit, the following constraint equation can be obtained:

[0177]

number

[0178]

number

[0179] (outside 65) JPEG0007639260000087.jpg8170 is the motion acceleration of the leg wheel (virtual single wheel), (outside 66) JPEG0007639260000088.jpg8170 is the motion acceleration of the left leg wheel part, (outside 67) JPEG0007639260000089.jpg9170 is the motion acceleration of the right leg wheel part, (outside 68) JPEG0007639260000090.jpg10170 is the target relative yaw angular acceleration, (outside 69) JPEG0007639260000091.jpg7170 is the relative distance between the left leg wheel part and the right leg wheel part along a direction perpendicular to the forward direction of the robot. Furthermore, expressions for the linear acceleration of the left leg wheel part and the right leg wheel part can be obtained from the relationship between the linear acceleration and the angular acceleration, and the meaning of each parameter amount can be referred to above.

[0180]

number

[0181]

number

[0182] (outside 70) JPEG0007639260000094.jpg10170 is the target motion acceleration of the left leg wheel part, (71 outside) JPEG0007639260000095.jpg11170 is the target motion acceleration of the right leg wheel part, (outside 72) JPEG0007639260000096.jpg12170 is the target motion acceleration of the leg wheel part (virtual single wheel), (outside 73) JPEG0007639260000097.jpg12170 is the target relative yaw angular acceleration of the leg wheel part (virtual single wheel), (outside 74) JPEG0007639260000098.jpg12170 is the relative distance between the left leg wheel part and the right leg wheel part at the current timing along a direction perpendicular to the forward direction of the robot.

[0183] By performing the calculation in this manner, the target acceleration of the left leg wheel part and the target acceleration of the right leg wheel part can be obtained. By installing in this manner, in order to realize balance control of the leg-wheel type robot, the left and right leg wheel parts are controlled so as to have a relative distance along the forward motion direction of the robot (for example, in the forward motion direction, the left leg wheel part is provided in front of the right leg wheel part) based on the actual terrain and motion situation.

[0184] For example, the leg-wheel robot is a dual-wheel robot with a parallel-connected leg configuration. In this case, the process of setting the constraint function of the leg-wheel robot is described in more detail as follows:

[0185] 7 is a three-dimensional schematic diagram of a robot 200A according to an embodiment of the present disclosure. Referring to FIG. 7, in some embodiments, the leg wheel unit 210A of the robot 200A includes a left leg wheel unit 211A and a right leg wheel unit 212A, each of the left leg wheel unit 211A and the right leg wheel unit 212A includes a single wheel unit, and each of the left leg wheel unit 211A and the right leg wheel unit 212A of the robot is further provided with two parallel-connected thigh units that are connected to the central axis of the wheel unit and realize the motion control of the wheel unit.

[0186] Referring to FIG. 7, the left leg wheel section 211A includes, for example, a left wheel 2110A and is further provided with a first left thigh 2111A and a second left thigh 2112A connected in parallel for controlling the left leg wheel, and the right leg wheel section 212A includes, for example, a right wheel 2120A and is further provided with a first right thigh 2121A and a second right thigh 2122A connected in parallel for controlling the right leg wheel.

[0187] In this case, in the process of setting the constraint function of the robot, a dynamics model of the robot is first established to establish the constraint relationship between the acceleration required for the control of each subtask (leg-wheel balance task, base rollover task) and the actual input joint moment. This is because the base of the leg-wheel robot is not fixedly connected to the inertial system, and there is a closed link in the kinematics.

[0188] In addition, if we assume that there is no sliding friction between the wheels and the ground and only pure rolling friction, there is a constraint that the acceleration of the contact point in the x direction (the x direction is the forward direction of the robot) and z direction (the z direction is the vertical upward direction) in the local coordinate system is 0. By calculating the Jacobian matrix for the general joint movement speed of the x and z direction velocities in the local coordinate system of the left and right leg wheel parts, the following equation can be obtained based on the acceleration constraint.

[0189]

number

[0190] α1 is the acceleration of the leg wheel part, (outside 75) JPEG0007639260000100.jpg10170 and (outside 76) JPEG0007639260000101.jpg12170 is the velocity of the left leg wheel in the x direction, (outside 77) JPEG0007639260000102.jpg12170 is the velocity of the left leg wheel in the z direction, (outside 78) JPEG0007639260000103.jpg10170 is the velocity of the right leg wheel in the x direction, (outside 79) JPEG0007639260000104.jpg7170 is the velocity of the right leg-wheel in the z direction, is the joint position of each joint of the leg-wheel type robot, (outside 80) JPEG0007639260000105.jpg10170 is the joint velocity of each joint of the robot, (Outside 81) JPEG0007639260000106.jpg8170 is the angular acceleration of each joint of the leg-wheel type robot, (Outside 82) JPEG0007639260000107.jpg8170 includes the leg-wheel joint angular acceleration and the base joint angular acceleration of the leg-wheel type robot, (Outside 83) JPEG0007639260000108.jpg9170 indicates that the value of the joint angular acceleration of the left and right leg-wheel parts of the leg-wheel robot is the target joint angular acceleration reference value v of the target robot joint of the leg-wheel robot in the above formula 3, and J c is the general joint motion velocity of the x-direction and z-direction velocity v1 of the left and right leg wheels in the local coordinate system. (Outside 84) JPEG0007639260000109.jpg8170 is the Jacobian matrix for

[0191] By performing calculations, the contact point acceleration offset for the general joint angular acceleration can be obtained.

[0192]

number

[0193] f α is an acceleration function, Contact is a parameter quantity related to the contact process, and the meanings of the other parameters can be found above.

[0194] Using a similar method, the parallel-connected thigh closed link constraint equations in the wide open link coordinate system are as follows:

[0195]

number

[0196] α2 is the acceleration of the parallel-connected leg, v2 is the velocity of the parallel-connected leg, and J λ is the general joint motion velocity of the parallel joint joint in the general open link coordinate system v2 (Outside 85) JPEG0007639260000112.jpg11170 is the Jacobian matrix for (Outside 86) JPEG0007639260000113.jpg8170 is the joint velocity of each joint of the robot, (Outside 87) JPEG0007639260000114.jpg9170 is the angular acceleration of each joint of the robot.

[0197] The acceleration offset is calculated by the following formula, where Close represents a parameter amount related to the close link constraint of the parallel-connected leg section, and the meanings of the other parameters can be seen above.

[0198]

number

[0199] In this way, the dynamic equations and the contact constraint equations 29 to 32 without external force interference are obtained.

[0200]

number

[0201]

number

[0202]

number

[0203]

number

[0204] The known quantities are M, C, G, S, and J. c , J λ , (Outside 88) JPEG0007639260000120.jpg12170 (Outside 89) JPEG0007639260000121.jpg13170 Specifically, the mass matrix M can be calculated quickly using the composite rigid body algorithm CRBA based on the joint position q, and the centrifugal and Coriolis force offset terms C are calculated based on the joint position q and the joint velocity (Outside 90) JPEG0007639260000122.jpg12170 can be quickly calculated using the Newton-Euler iterative algorithm RNEA based on the gravity offset term G, can be quickly calculated using the Newton-Euler iterative algorithm RNEA based on the joint position q, and the selection matrix S is used to distinguish between active and non-active joints, (Outside 91) JPEG0007639260000123.jpg14170 It is written as a i = 1 indicates that it is an unactuated joint, and the first four terms correspond to the joints of the base, which can be directly set to 1 since the base is unactuated.

[0205] The variables are (Outside 92) JPEG0007639260000124.jpg10170 It is. (Outside 93) JPEG0007639260000125.jpg12170 is the joint angular acceleration, τ is the moment term corresponding to the joint, and F c is the generalized force at the contact point in the local coordinate system. Specifically, (Outside 94) JPEG0007639260000126.jpg15170 and (Outside 95) JPEG0007639260000127.jpg13170 is the generalized force in the x direction of the left leg-wheel of the leg-wheel type robot, (outside 96) JPEG0007639260000128.jpg14170 is the generalized force in the z direction of the left leg-wheel part of the leg-wheel type robot, (Outside 97) JPEG0007639260000129.jpg14170 is the generalized force in the x direction of the right leg-wheel of the leg-wheel type robot, (Outside 98) JPEG0007639260000130.jpg13170 is the generalized force in the z direction of the right leg-wheel part of the leg-wheel type robot. F λ is the closing link force acting on the rear leg from the front leg, (outside 99) JPEG0007639260000131.jpg14170 and (outside 100) JPEG0007639260000132.jpg10170 is the force acting in the x direction from the front leg of the two parallel-connected legs of the left leg wheel part on the rear leg, (Outside 101) JPEG0007639260000133.jpg10170 is the force acting in the z direction on the rear leg from the front leg of the two parallel-connected legs of the left leg wheel unit, (Outside 102) JPEG0007639260000134.jpg10170 is the force acting in the x direction on the rear leg of the right leg wheel unit, of the two parallel-connected legs of the front leg, (Outside 103) JPEG0007639260000135.jpg12170 is the force acting in the z direction on the rear leg of the right wheel unit, which is one of the two parallel-connected legs.

[0206] By adopting the full dynamics method, the above equality constraints can be obtained, and the contact force F in the local system (in the direction of the friction curve tangent line) can be calculated based on the friction cone constraint. c , the constraints that must be satisfied can be obtained.

[0207]

number

[0208]

number

[0209] (Outside 104) JPEG0007639260000138.jpg12170 is the generalized force of the left leg-wheel of the leg-wheel type robot in the x direction, μ is the friction factor in the current surrounding environment, (Outside 105) JPEG0007639260000139.jpg13170 is the generalized force in the z direction of the right leg-wheel of the leg-wheel type robot, (Outside 106) JPEG0007639260000140.jpg13170 is the generalized force in the x direction of the right leg-wheel of the leg-wheel type robot, (Outside 107) JPEG0007639260000141.jpg10170 is the generalized force in the z direction of the right leg-wheel part of the leg-wheel type robot.

[0210] Obtain the friction cone constraint matrix:

[0211]

number

[0212] (Outside 108) JPEG0007639260000143.jpg9170 is the friction constraint matrix corresponding to the left wheel contact point of the leg-wheeled robot, (Outside 109) JPEG0007639260000144.jpg10170 is the friction constraint matrix corresponding to the right wheel contact point of the leg-wheeled robot, and μ is the friction factor in the current surrounding environment.

[0213] In this manner, an equality constraint sub-function, an inequality constraint sub-function, and a threshold constraint sub-function of the constraint function are obtained.

[0214] The equality constraint sub-functions include:

[0215]

number

[0216] (Outside 110) JPEG0007639260000146.jpg11170 is the joint angular acceleration of the leg-wheel type robot at the target timing, (Outside 111) JPEG0007639260000147.jpg13170 is the current joint velocity of the leg-wheel robot, (Outside 112) JPEG0007639260000148.jpg10170 is the generalized force of the contact point in the local coordinate system of the leg-wheeled robot at the target timing, (Outside 113) JPEG0007639260000149.jpg11170 is the force acting between the two parallel-connected legs at the target timing, (Outside 114) JPEG0007639260000150.jpg10170 is the joint moment of the leg-wheel type robot at the target timing, (Outside 115) JPEG0007639260000151.jpg11170 , (Outside 116) JPEG0007639260000152.jpg9170 is the generalized force F of the contact point in the local coordinate system of the leg-wheel type robot at the current timing. c , and the acting force F between the two parallel connected thighs λ are generated based on the current motion state data of the leg-wheel type robot, and C t , G t , M t is a parameter amount generated based on the current joint movement state data of the leg-wheel type robot, S is a parameter amount determined based on the joint configuration of the leg-wheel type robot, I is an identity matrix, the number of dimensions of which is determined based on the number of degrees of freedom of movement of the leg-wheel type robot, t is the current timing of the movement of the leg-wheel type robot, and t+1 is the target timing of the movement of the leg-wheel type robot.

[0217] The joint inequality constraint sub-functions include:

[0218]

number

[0219] (Outside 117) JPEG0007639260000154.jpg8170 is the friction constraint matrix corresponding to the left wheel contact point of the leg-wheeled robot, (Outside 118) JPEG0007639260000155.jpg7170 is the friction constraint matrix corresponding to the right wheel contact point of the leg-wheeled robot. The friction cone constraint matrix is ​​specifically expressed as follows:

[0220]

number

[0221] The joint threshold constraint sub-functions include:

[0222]

number

[0223] In the above equality and inequality constraints, the variables (Outside 119) JPEG0007639260000158.jpg9170 There is some correlation between the input moments, i.e., (Outside 120) JPEG0007639260000159.jpg8170 is known and there are no inequality constraints, (Outside 121) JPEG0007639260000160.jpg9170 A least squares solution can be found for

[0224]

number

[0225] In this way, (Outside 122) JPEG0007639260000162.jpg24170 belongs to a high-dimensional space, but in fact has a one-to-one mapping relationship with a joint moment subspace whose dimension is k (k is much smaller than the dimension of the high-dimensional space), i.e., there are equality and inequality constraints, (Outside 123) JPEG0007639260000163.jpg20170 actually belongs to a low-dimensional bounded manifold that satisfies the equality constraints in the high-dimensional space. c By rationally choosing , we can obtain the τ values ​​corresponding to the moment subspace.

[0226] In addition, the linear mapping relationship between the operational space acceleration and the general joint space acceleration is defined by the Jacobian matrix, and the joint moment input that satisfies the dynamics constraints can be found by finding the operational space acceleration that matches the low-dimensional manifold degrees of freedom. Furthermore, the control within the low-dimensional manifold for the operational space is realized (the new control needs of the robot do not destroy the equality and inequality constraints).

[0227] In addition, for the selection of the operation space, the final total task Jacobian matrix (Outside 124) JPEG0007639260000164.jpg13170 but (Outside 125) JPEG0007639260000165.jpg10170 (where k is the dimensionality of the joint space) is satisfied. (Outside 126) JPEG0007639260000166.jpg9170 We show that if , then there exists an uncontrollable subspace that causes the dynamical system to diverge. (Outside 127) JPEG0007639260000167.jpg8170 If so, this indicates that there may be task needs that violate the constraints. (Outside 128) JPEG0007639260000168.jpg8170 Then, for example, a final criterion can be obtained.

[0228] Based on the above, when the above dynamics equations and friction constraints are satisfied, a correspondence relationship can be determined between the operation space of the leg-wheel robot (where robot motion control is performed) and the joint motion space of the leg-wheel robot (where each joint of the leg-wheel robot performs a corresponding joint motion). In this way, in the operation space of the leg-wheel robot, the target motion acceleration of the leg-wheel robot determined based on the leg-wheel target motion acceleration and the base target motion acceleration is transformed into the joint motion space of the leg-wheel robot through the correspondence relationship. Similarly, the target motion acceleration of the leg-wheel robot can also be represented by the joint angular acceleration in the joint motion space of the leg-wheel robot.

[0229] FIG. 8 is a schematic diagram of a control frame of a leg-wheel robot provided by one exemplary embodiment of the present application.

[0230] For example, the motion state data of the leg-wheel robot 100 is acquired, and the motion state data includes the pitch angle and angular velocity of the base, the angles and angular velocities of the joints, etc. The motion state data is sent to the state estimation module 201, which estimates the system internal state according to the motion state data. The output of the state estimation module 201 is input to the nonlinear controller 202, the walking motion trajectory generator 204, and the multi-stage jump trajectory optimizer 205.

[0231] For example, the nonlinear controller 202 performs nonlinear control for the leg-wheeled robot to obtain a target joint angular acceleration reference value of a target robot joint of the leg-wheeled robot, which may be used as a reference input for the wheel balance task in the whole body dynamics controller 206.

[0232] For example, the user input module 203, the walking motion trajectory generator 204, and the multi-stage jump trajectory optimizer 205 obtain a motion reference trajectory of the leg-wheel robot. The motion reference trajectory obtained by the user input module 203 can be the reference input of the leg-wheel balance task, the wheel roll task, the base height task, the base roll task, the base pitch task, and the tail task of the whole-body dynamics controller 206, the motion reference trajectory obtained by the walking motion trajectory generator 204 can be the reference input of the walking task of the whole-body dynamics controller 206, and the motion reference trajectory obtained by the multi-stage jump trajectory optimizer 205 can be the reference input of the jump task of the whole-body dynamics controller 206.

[0233] For example, the whole body dynamics controller 206 obtains and calculates the target joint angular acceleration reference value and the motion reference trajectory of the target robot joint of the leg-wheeled robot to obtain the joint moment for controlling the leg-wheeled robot to perform the control task, the joint moment including the moment of each joint motor of the leg-wheeled robot. The moment output from the whole body dynamics controller 206 is used as the input of the leg-wheeled robot 100 to instruct the leg-wheeled robot 100 to complete the control task.

[0234] FIG. 9 is a structural block diagram of a control device of a leg-wheel type robot provided in one exemplary embodiment of the present application. As shown in FIG. 9, the device includes: An acquisition module 920 for acquiring current motion state data of the leg-wheel robot, where the current motion state data indicates motion characteristics of the leg-wheel robot; A decision module 940 inputs the current motion state data into a nonlinear controller to obtain a target joint angular acceleration reference value of a target robot joint of the leg-wheeled robot; and an output module 960 that inputs the target joint angular acceleration reference value to the whole body dynamics controller to output joint moments for controlling the leg-wheeled robot to perform the control task.

[0235] In one possible embodiment, the nonlinear controller is obtained by the steps of: determining an original energy relationship for the leg-wheeled robot, where the original energy relationship includes an original potential energy, and the original potential energy has a non-stationary characteristic; determining a variable energy relationship for the leg-wheeled robot, where the variable energy relationship includes a variable potential energy, and the variable potential energy has a stationary characteristic, and when the leg-wheeled robot is in a target balance state, the variable potential energy value is minimum; and determining a nonlinear controller with the original energy relationship change as a target being the variable energy relationship.

[0236] In one possible embodiment, the nonlinear controller is (Outside 129) JPEG0007639260000169.jpg11170 where v is the target joint angular acceleration reference value of the target robot joint, x is the rotation distance of the leg-wheel part of the leg-wheel type robot, (Outside 130) JPEG0007639260000170.jpg9170 is the linear velocity of the leg-wheel part of the leg-wheel type robot, and x * indicates the balance point of the leg-wheel assembly in the target balance state, (Outside 131) JPEG0007639260000171.jpg12170 θ is the inclination angle of the leg-wheel robot, (Outside 132) JPEG0007639260000172.jpg8170 denotes the tilt angular velocity of the leg-wheel robot, and k v are free parameters in the decay injection process, and A1, A2, A3 and A4 are preset parameters.

[0237] In one possible embodiment, the nonlinear controller is (Outside 133) JPEG0007639260000173.jpg10170 v denotes the target joint angular acceleration reference value of the target robot joint; (Outside 134) JPEG0007639260000174.jpg12170 and (Outside 135) JPEG0007639260000175.jpg9170 indicates the rotational linear velocity of the leg-wheel part of the leg-wheel type robot, (Outside 136) JPEG0007639260000176.jpg9170 is the rotational acceleration of the leg-wheel robot in the target balance state, and θ is the inclination angle of the leg-wheel robot. (Outside 137) JPEG0007639260000177.jpg8170 denotes the tilt angular velocity of the leg-wheel robot, and k v are free parameters in the decay injection process, and A2, A3 and A4 are preset parameters.

[0238] In one possible embodiment, the output module 960 solves a quadratic optimization problem using the target joint angular acceleration reference values ​​as control targets by a whole body dynamics controller, subject to satisfying constraints, to obtain joint moments for controlling the leg-wheeled robot to perform a self-balancing task.

[0239] In one possible embodiment, the output module 960 calculates a target attitude angular acceleration reference value of the base unit based on the target joint angular acceleration reference value of the leg-wheel unit with the target of maintaining relative balance, and determines, by the whole body dynamics controller, joint moments of each joint when the leg-wheel robot performs a self-balancing task based on the target joint angular acceleration reference value of the leg-wheel unit and the target attitude angular acceleration reference value of the base unit under the premise that the constraint condition is satisfied. In one possible embodiment, the acquisition module 920 further acquires a motion reference trajectory of the motion task, and the output module 960 inputs the target joint angular acceleration reference value and the motion reference trajectory into the whole body dynamics controller to output joint moments for controlling the leg-wheel robot to perform the motion task.

[0240] In one possible embodiment, the output module 960 solves a quadratic optimization problem using the target joint angular acceleration reference value and the motion reference trajectory as control targets by the whole body dynamics controller, subject to satisfying constraints, to obtain joint moments for controlling the leg-wheeled robot to perform a motion task.

[0241] In one possible embodiment, the output module 960 calculates a target attitude angular acceleration reference value of the base unit by targeting the motion reference trajectory based on the target joint angular acceleration reference value of the leg-wheel unit, and, assuming that the constraint conditions are satisfied, determines, by the whole-body dynamics controller, the joint moment of each joint when the leg-wheel robot performs the motion task based on the target joint angular acceleration reference value of the leg-wheel unit and the target attitude angular acceleration reference value of the base unit.

[0242] In one possible embodiment, the acquisition module 920 further receives a user control input from a remote control, and determines a motion reference trajectory of the motion task based on the user control input.

[0243] In one possible embodiment, the leg-wheel robot includes a leg-wheel section and a base section, and the movement task includes at least one of a rollover task of the wheels of the leg-wheel robot, a height task of the base section of the leg-wheel robot, and a balance task of the leg-wheel section.

[0244] In one possible embodiment, the acquisition module 920 further inputs the current motion state data into a trajectory generator of the motion task to obtain a motion reference trajectory of the motion task.

[0245] In one possible embodiment, the acquisition module 920 further inputs the current motion state data into a walking motion trajectory generator to obtain a motion reference trajectory for controlling the leg-wheeled robot to perform a dual-wheel alternating walking task.

[0246] In one possible embodiment, the acquisition module 920 further inputs the current motion state data into a jump trajectory generator to obtain a motion reference trajectory for controlling the leg-wheel robot to perform a jump task.

[0247] In one possible embodiment, the constraints include at least two of a friction cone constraint, a closed loop link constraint, and a dynamics constraint; The friction cone constraint is used to constrain the load-bearing surface reaction forces to satisfy the friction cone, where the load-bearing surface reaction forces are the acting forces of the load-bearing surface on the leg-wheeled robot, the closed-loop link constraint is used to constrain the thigh link structure of the leg-wheeled robot, and the dynamics constraint is used to constrain the control task based on the dynamics model of the leg-wheeled robot.

[0248] In one possible implementation, when the wheels of the leg-wheeled robot are on the ground, the constraints include a friction cone constraint, a closed-loop link constraint, and a dynamics constraint, and when the wheels of the leg-wheeled robot are not on the ground, the constraints include a closed-loop link constraint and a dynamics constraint.

[0249] In one possible embodiment, the acquisition module 920 acquires sensor data of the leg-wheel type robot, estimates a motion state of the leg-wheel type robot according to the sensor data, and acquires current motion state data; The sensor data includes at least one of the left wheel angle and left wheel angular velocity of the leg-wheel robot, at least one of the right wheel angle and right wheel angular velocity of the leg-wheel robot, the pitch angle and angular velocity of the base part, the roll angle and angular velocity of the base part, the yaw angle and angular velocity of the base part, and at least one of the angle and angular velocity of the joint motor of the leg-wheel robot.

[0250] 10 is a structural schematic block diagram of a leg-wheeled robot provided by an embodiment of the present application. The leg-wheeled robot in this embodiment of FIG. 10 includes one or more processors 801, one or more sensors 802, one or more motors 803, and a memory 804. The above-mentioned processor 801, sensor 802, motor 803, and memory 804 are connected via a bus 805. The memory 804 stores a computer program, the computer program includes program instructions, and the processor 801 executes the program instructions stored in the memory 804.

[0251] The memory 804 is a volatile memory, such as a random-access memory. The memory may include non-volatile memory, such as flash memory, solid-state drive, etc. The memory may include a variety of storage devices, such as a hard disk drive, a hard disk drive (HDD), ...

[0252] The processor 801 may be a central processing unit (CPU). The processor 801 further includes a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC). The PLD may be a field-programmable gate array (FGA), a programmable logic device (PLD), or the like. The processor 801 may be a FPGA (Field Gate Array), a Generic Array Logic (GAL), etc. The processor 801 may be a combination of the above structures.

[0253] In an embodiment of the present application, the memory 804 stores a computer program, the computer program includes program instructions, and the processor 801 executes the program instructions stored in the memory 804 to realize the corresponding method steps of Figures 3 to 5 above.

[0254] In one embodiment, the processor 801 executes a program command to: Inputting the current motion state data into a nonlinear controller to obtain a target joint angular acceleration reference value of a target robot joint of the leg-wheel robot; inputting a target joint angular acceleration reference value of a target robot joint of the leg-wheeled robot to a whole-body dynamics controller to output a joint moment for controlling the leg-wheeled robot to perform the control task.

[0255] In one embodiment, the nonlinear system controller includes a feedback linearization controller and a nonlinear controller based on a simplified model, and the processor 801 calls a program instruction to: determining an original energy relationship of the leg-wheel robot, the original energy relationship including an original potential energy, the original potential energy having a non-stationary characteristic; determining a variable energy relationship of the leg-wheel robot, the variable energy relationship including a variable potential energy, the variable potential energy having a steady-state characteristic, and when the leg-wheel robot is in a target balance state, the variable potential energy value is a minimum; determining a nonlinear controller targeting the original energy relation change to be the variable energy relation.

[0256] In one embodiment, the control operation includes a touch operation by a game user on a user interface, and the processor 801 invokes a program instruction to: The whole-body dynamics controller is configured to perform the steps of solving a quadratic optimization problem using the target joint angular acceleration reference value as a control target, provided that the constraint conditions are satisfied, to obtain joint moments for controlling the leg-wheeled robot to perform a self-balancing task.

[0257] In one embodiment, the processor 801 executes a program command to: calculating a target attitude angular acceleration reference value of the base unit based on the target joint angular acceleration reference value of the leg-wheel unit, with a target of maintaining relative balance; The system is configured to execute a step of determining, by the whole-body dynamics controller, joint moments of each joint when the leg-wheel robot performs a self-balancing task based on the target joint angular acceleration reference value of the leg-wheel section and the target attitude angular acceleration reference value of the base section, provided that the constraint conditions are satisfied.

[0258] In one embodiment, the processor 801 executes a program command to: The system is arranged to perform the steps of inputting the target joint angular acceleration reference values ​​and the motion reference trajectory to a whole body dynamics controller to output joint moments for controlling the leg-wheeled robot to perform the motion task.

[0259] In one embodiment, the processor 801 executes a program command to: Provided that the constraint conditions are satisfied, the whole-body dynamics controller is configured to execute a step of solving a quadratic optimization problem using the target joint angular acceleration reference value and the motion reference trajectory as control targets to obtain joint moments when controlling the leg-wheeled robot to perform a motion task.

[0260] In one embodiment, the processor 801 executes a program command to: calculating a target posture angular acceleration reference value of the base unit based on the target joint angular acceleration reference value of the leg-wheel unit, with the motion reference trajectory as a target; The system is configured to execute a step of determining, by the whole-body dynamics controller, joint moments of each joint when the leg-wheel robot performs an operation task based on the target joint angular acceleration reference value of the leg-wheel section and the target attitude angular acceleration reference value of the base section, provided that the constraint conditions are satisfied.

[0261] In one embodiment, the processor 801 executes a program command to: The device is arranged to perform the steps of inputting the current motion state data into a trajectory generator for the motion task, and obtaining a motion reference trajectory for the motion task.

[0262] In one embodiment, the processor 801 executes a program command to: The system is arranged to perform the steps of inputting the current motion state data into a walking motion trajectory generator to obtain a motion reference trajectory for controlling the leg-wheeled robot to perform the dual-wheel alternating walking task.

[0263] In one embodiment, the processor 801 executes a program command to: The system is arranged to perform the steps of inputting the current motion state data into a jump trajectory generator to obtain a motion reference trajectory for controlling the leg-wheeled robot to perform a jump task.

[0264] The sensor 802 acquires motion state data of the leg-wheel robot, and the motion state data includes at least one of the following: the left wheel speed of the leg-wheel robot, the right wheel speed of the leg-wheel robot, the pitch angle and angular velocity of the base part of the leg-wheel robot, the roll angle and angular velocity of the base part of the leg-wheel robot, the yaw angle and angular velocity of the base part of the leg-wheel robot, and the angle and angular velocity of the joint motor of the leg-wheel robot.

[0265] The motor 803 controls the motion of the leg-wheel robot to complete the motion task, and includes each joint motor and wheel motor of the leg-wheel robot.

[0266] An embodiment of the present application further provides a leg-wheeled robot, which includes a processor and a memory, wherein at least one program is stored in the memory, and the at least one program is loaded and executed by the processor to realize the control method for the leg-wheeled robot provided by each of the above method embodiments.

[0267] An embodiment of the present application further provides a computer device, the computer device including a processor and a memory, at least one program stored in the memory, the at least one program being loaded and executed by the processor to realize the control method for a leg-wheeled robot provided by each of the above method embodiments.

[0268] An embodiment of the present application further provides a computer-readable storage medium, in which at least one program is stored, and the at least one program is loaded and executed by a processor to realize the control method for a leg-wheeled robot provided by each of the above method embodiments.

[0269] Optionally, the computer readable storage medium comprises a read only memory (ROM, Read Only Memory Random access memory includes resistive random access memory (ReRAM) and dynamic random access memory (DRAM). The numbers of the embodiments in the present application above do not represent the superiority or inferiority of the embodiments, but are merely for the purpose of explanation.

[0270] As can be understood by those skilled in the art, the realization of all or part of the steps of the above embodiments may be completed by hardware, or may be instructed to complete the steps by using a program, and the above-mentioned program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a magnetic disk, an optical disk, etc. The above is not limiting the present application, but is only a preferred embodiment of the present application, and any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application fall within the protection scope of the present application.

[0271] The embodiments of the present application further provide a chip, which includes a programmable logic circuit and a program, and the chip is used to realize the control method for a leg-wheeled robot provided by each of the above method embodiments.

[0272] An embodiment of the present application further provides a computer program product, the computer program product including computer instructions stored in a computer-readable storage medium, a processor of a computing device reading and executing the computer instructions from the computer-readable storage medium, thereby causing the computing device to perform the method for controlling a leg-wheeled robot according to any one of the above embodiments.

[0273] The above is not intended to limit the present application, but is merely a preferred embodiment of the present application, and any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application are within the scope of protection of the present application.

Claims

1. A method for controlling a leg-wheel type robot executed by a processor of the leg-wheel type robot, comprising: acquiring current motion state data of the leg-wheel type robot, the current motion state data indicating motion characteristics of the leg-wheel type robot; inputting the current motion state data into a nonlinear controller to obtain a target joint angular acceleration reference value of a target robot joint of the leg-wheeled robot; inputting the target joint angular acceleration reference values ​​into a whole-body dynamics controller to output joint moments for controlling the leg-wheeled robot to perform a control task; Including, The nonlinear controller is determining an original energy relation of the leg-wheel robot, the original energy relation including an original potential energy, the original potential energy having a non-stationary characteristic; determining a variable energy relationship of the leg-wheel robot, the variable energy relationship including a variable potential energy, the variable potential energy having a steady-state characteristic, and a value of the variable potential energy being minimum when the leg-wheel robot is in a target balance state; determining the nonlinear controller with the original energy relationship change as a target to be the variable energy relationship; The step of inputting the current motion state data into a nonlinear controller to obtain a target joint angular acceleration reference value of a target robot joint of the leg-wheel robot includes: and calculating a target joint angular acceleration reference value of a target robot joint of the leg-wheeled robot by substituting the current motion state data into a nonlinear controller represented by the following equation: [0010] v denotes a target joint angular acceleration reference value of the target robot joint, x denotes a rotation distance of a leg-wheel portion of the leg-wheel type robot, (Outside 1) 【number】 represents the rotational linear velocity of the leg-wheel portion of the leg-wheel type robot, (Outer 2) 【number】 indicates a balance point of the leg-wheel unit in the target balance state, (Outer 3) 【number】 θ is the inclination angle of the leg-wheel robot, (outer 4) 【number】 represents the tilt angular velocity of the leg-wheel robot, k v is a free parameter in the damping injection process, and A 1 , A 2 , A 3 and A 4 are pre-established parameter calculation expressions.

2. A method for controlling a leg-wheeled robot executed by a processor of the leg-wheeled robot, comprising: acquiring current motion state data of the leg-wheel type robot, the current motion state data indicating motion characteristics of the leg-wheel type robot; inputting the current motion state data into a nonlinear controller to obtain a target joint angular acceleration reference value of a target robot joint of the leg-wheeled robot; inputting the target joint angular acceleration reference values ​​into a whole-body dynamics controller to output joint moments for controlling the leg-wheeled robot to perform a control task; Including, The nonlinear controller is determining an original energy relation of the leg-wheel robot, the original energy relation including an original potential energy, the original potential energy having a non-stationary characteristic; determining a variable energy relationship of the leg-wheel robot, the variable energy relationship including a variable potential energy, the variable potential energy having a steady-state characteristic, and a value of the variable potential energy being minimum when the leg-wheel robot is in a target balance state; determining the nonlinear controller with the original energy relationship change as a target to be the variable energy relationship; The step of inputting the current motion state data into a nonlinear controller to obtain a target joint angular acceleration reference value of a target robot joint of the leg-wheel robot includes: and calculating a target joint angular acceleration reference value of a target robot joint of the leg-wheeled robot by substituting the current motion state data into a nonlinear controller represented by the following equation: [0025] v denotes the target joint angular acceleration reference value of the target robot joint, (outside 5) 【number】 and (outer 6) 【number】 represents the rotational linear velocity of the leg-wheel portion of the leg-wheel type robot, (Outer 7) 【number】 represents the rotational acceleration of the leg-wheel part of the leg-wheel type robot in the target balance state, θ represents the inclination angle of the leg-wheel type robot, (Outside 8) 【number】 represents the tilt angular velocity of the leg-wheel type robot, and k ｖ is a free parameter in the decay injection process, and A ２ , A ３ and A ４ is a preset parameter calculation expression.

3. the control task includes a self-balancing task; inputting the target joint angular acceleration reference values ​​into a whole body dynamics controller to output joint moments for controlling the leg-wheeled robot to perform a control task, 3. The method of claim 1, further comprising: solving a quadratic optimization problem by the whole-body dynamics controller using the target joint angular acceleration reference values ​​as control targets, subject to satisfaction of constraints, to obtain joint moments for controlling the leg-wheeled robot to perform the self-balancing task.

4. the leg-wheel type robot includes a leg-wheel section and a base section, and a target joint angular acceleration reference value of the target robot joint includes a target angular acceleration reference value of the leg-wheel section; solving a quadratic optimization problem by the whole-body dynamics controller using the target joint angular acceleration reference value as a control target to obtain joint moments for controlling the leg-wheeled robot to perform the self-balancing task, assuming that the constraint conditions are satisfied, calculating a target attitude angular acceleration reference value of the base unit based on the target joint angular acceleration reference value of the leg-wheel unit, with a target of maintaining relative balance; and determining, by the whole-body dynamics controller, joint moments of each joint when the leg-wheel robot performs the self-balancing task based on the target joint angular acceleration reference value of the leg-wheel section and the target attitude angular acceleration reference value of the base section, assuming that the constraint conditions are satisfied.

5. The control task includes an operation task, and the method further comprises: obtaining a motion reference trajectory for the motion task; Further comprising: inputting the target joint angular acceleration reference values ​​into a whole body dynamics controller to output joint moments for controlling the leg-wheeled robot to perform a control task, 3. The method of claim 1 or 2, comprising inputting the target joint angular acceleration reference values ​​and the motion reference trajectory into the whole body dynamics controller to output joint moments for controlling the leg-wheeled robot to perform the motion task.

6. inputting the target joint angular acceleration reference values ​​and the motion reference trajectory into the whole body dynamics controller to output joint moments for controlling the leg-wheeled robot to perform the motion task, 6. The method of claim 5, further comprising: solving a quadratic optimization problem by the whole body dynamics controller using the target joint angular acceleration reference value and the motion reference trajectory as control targets, provided that constraint conditions are satisfied, to obtain the joint moments when controlling the leg-wheeled robot to perform the motion task.

7. the leg-wheel type robot includes a leg-wheel section and a base section, and a target joint angular acceleration reference value of the target robot joint includes a target angular acceleration value of the leg-wheel section; a step of solving a quadratic optimization problem by the whole-body dynamics controller using the target joint angular acceleration reference value and the motion reference trajectory as control targets to obtain the joint moments when controlling the leg-wheeled robot to perform the motion task, on the premise that the constraint conditions are satisfied, calculating a target attitude angular acceleration reference value of the base unit based on the target joint angular acceleration reference value of the leg-wheel unit, with the motion reference trajectory as a target; and determining, by the whole body dynamics controller, joint moments of each joint when the leg-wheel type robot performs the movement task based on the target joint angular acceleration reference value of the leg-wheel section and the target attitude angular acceleration reference value of the base section, assuming that the constraint conditions are satisfied.

8. The step of acquiring a motion reference trajectory for the motion task includes: receiving a user control input from a remote control; and determining a motion reference trajectory for the motion task based on the user control input.

9. The leg-wheel type robot includes a leg-wheel unit and a base unit, and the operation task includes: A rollover task of the leg wheel unit; A balancing task for the leg-wheel unit; a base height task; A base portion rollover task; A pitching task of the base portion; A yaw task of the base portion; a tail control task of the base; and a balancing task for the leg wheel assembly.

10. The step of acquiring a motion reference trajectory for the motion task includes: The method of claim 5 , further comprising inputting the current motion state data into a trajectory generator for the motion task to obtain a motion reference trajectory for the motion task.

11. The trajectory generator includes a walking motion trajectory generator; The step of inputting the current motion state data into a trajectory generator to obtain a motion reference trajectory for the motion task includes:

11. The method of claim 10, comprising inputting the current motion state data into the walking motion trajectory generator to obtain a motion reference trajectory for controlling the leg-wheeled robot to perform a dual-wheel alternating walking task.

12. The trajectory generator includes a jump trajectory generator; The step of inputting the current motion state data into a trajectory generator to obtain a motion reference trajectory for the motion task includes:

11. The method of claim 10, comprising inputting the current motion state data into the jump trajectory generator to obtain a motion reference trajectory for controlling the leg-wheeled robot to perform a jump task.

13. the constraints include at least two of a friction cone constraint, a closed loop link constraint, and a dynamics constraint; 4. The method of claim 3, wherein the friction cone constraint is used to constrain load-bearing surface reaction forces to satisfy a friction cone, the load-bearing surface reaction forces being acting forces of a load-bearing surface on the leg-wheeled robot, the closed-loop link constraint is used to constrain a thigh link structure of the leg-wheeled robot, and the dynamics constraint is used to constrain the control task based on a dynamics model of the leg-wheeled robot.

14. When the wheels of the leg-wheeled robot land, the constraints include a friction cone constraint, a closed-loop link constraint, and a dynamics constraint; The method of claim 13 , wherein the constraints include the closed-loop link constraints and the dynamics constraints when the wheels of the leg-wheeled robot are not on the ground.

15. the leg-wheel type robot includes a leg-wheel unit and a base unit, the leg-wheel unit including a left wheel and a right wheel, The step of acquiring current motion state data of the leg-wheel type robot includes: acquiring sensor data of the leg-wheeled robot; estimating a motion state of the leg-wheeled robot based on the sensor data to obtain the current motion state data; Including, 3. The method of claim 1 or 2, wherein the sensor data includes at least one of a left wheel angle and a left wheel angular velocity of the leg-wheeled robot, at least one of a right wheel angle and a right wheel angular velocity of the leg-wheeled robot, a pitch angle and an angular velocity of the base part, a roll angle and an angular velocity of the base part, a yaw angle and an angular velocity of the base part, and an angle and an angular velocity of a joint motor of the leg-wheeled robot.

16. A control device for a leg-wheel type robot, comprising: an acquisition module for acquiring current motion state data of the leg-wheel type robot, the current motion state data indicating motion characteristics of the leg-wheel type robot; A decision module inputs the current motion state data into a nonlinear controller to obtain a target joint angular acceleration reference value of a target robot joint of the leg-wheeled robot; an output module inputting the target joint angular acceleration reference value to a whole body dynamics controller and outputting joint moments for controlling the leg-wheeled robot to perform a control task; Including, The nonlinear controller is determining an original energy relation of the leg-wheel robot, the original energy relation including an original potential energy, the original potential energy having a non-stationary characteristic; determining a variable energy relationship of the leg-wheel robot, the variable energy relationship including a variable potential energy, the variable potential energy having a steady-state characteristic, and a value of the variable potential energy being minimum when the leg-wheel robot is in a target balance state; determining the nonlinear controller with the original energy relationship change as a target to be the variable energy relationship; The step of inputting the current motion state data into a nonlinear controller to obtain a target joint angular acceleration reference value of a target robot joint of the leg-wheel robot includes: and calculating a target joint angular acceleration reference value of a target robot joint of the leg-wheeled robot by substituting the current motion state data into a nonlinear controller represented by the following equation: [0030] v denotes a target joint angular acceleration reference value of the target robot joint, x denotes a rotation distance of a leg-wheel portion of the leg-wheel type robot, (Outer 9) 【number】 represents the rotational linear velocity of the leg-wheel portion of the leg-wheel type robot, (Outside 10) 【number】 indicates a balance point of the leg-wheel unit in the target balance state, (Outside 11) 【number】 θ is the inclination angle of the leg-wheel robot, (outer 12) 【number】 denotes the tilt angular velocity of the leg-wheel robot, k v is a free parameter in the damping injection process, and A 1 , A 2 , A 3 and A 4 are pre-established parameter calculation expressions of the apparatus.

17. 1. A leg-wheeled robot including a processor and a memory, A leg-wheel type robot, wherein at least one program is stored in the memory, and the at least one program is loaded and executed by the processor to realize the control method for a leg-wheel type robot described in claim 1 or 2.

18. A computing device including a processor and a memory, At least one program is stored in the memory, and the at least one program is loaded and executed by the processor, thereby realizing the control method for a leg-wheeled robot described in claim 1 or 2.

19. A chip containing a programmable logic circuit or program, The chip is used to realize the control method for a leg-wheeled robot according to claim 1 or 2.

20. 3. A computer program including computer instructions, the computer instructions being stored in a computer-readable storage medium, and a processor reading and executing the computer instructions from the computer-readable storage medium, thereby realizing the method for controlling a leg-wheeled robot according to claim 1 or 2.

Images