Admittance control method and device of double-wheel-foot robot and double-wheel-foot robot

Abstract

The invention discloses an admittance control method and device of a double-wheel-foot robot and the double-wheel-foot robot, and is applied to the technical field of robotics.The method comprises the steps that admittance control is conducted based on initial expected leg length data of all legs and ground contact force, and target expected leg length data of all the legs are obtained; wherein the ground contact force is obtained by performing contact force estimation according to the target motor torque of each leg in the previous control period, and the ground contact force in the first control period is zero; performing torque processing by using the target expected leg length data, the actual leg length data and other control data of each leg to obtain a target motor torque of each leg; and controlling each leg according to the target motor torque of each leg. The double-wheel-foot robot can adapt to various complex terrains, the height stability of the head of the double-wheel-foot robot is kept, and therefore the control precision of the double-wheel-foot robot is effectively improved.

Description

Technical Field

[0001] This application relates to the technical field of robots, and in particular to an admittance control method, device and two-wheeled biped robot. Background Art

[0002] The two-wheeled biped robot combines the speed advantage of the wheeled robot and the maneuverability and obstacle-crossing ability of the legged robot, and can move efficiently and flexibly in various complex terrains, with broad application prospects. Currently, the two-wheeled biped robot usually ensures that it will not roll over during movement by fixing the expected leg length and the roll angle direction, thereby improving its stability during movement. However, in practical applications, the leg length of the two-wheeled biped robot will affect the stability of its head. When facing complex terrains such as slopes, the two-wheeled biped robot tends to maintain a fixed expected leg length through a proportional-integral-derivative controller, while ignoring the flexible control of its legs. This results in the two-wheeled biped robot being difficult to maintain the stability of its head under global control, and its legs being difficult to adapt to complex terrains, thus causing the control accuracy of the two-wheeled biped robot to decrease. Summary of the Invention

[0003] The embodiments of this application provide an admittance control method, device and two-wheeled biped robot for enabling the two-wheeled biped robot to adapt to various complex terrains and maintain the high stability of its head, thereby effectively improving the control accuracy of the two-wheeled biped robot.

[0004] On the one hand, the embodiments of this application provide an admittance control method for a two-wheeled biped robot, including the following steps:

[0005] Based on the initial expected leg length data of each leg and the ground contact force, perform admittance control to obtain the target expected leg length data of each leg; wherein, the ground contact force is obtained by estimating the contact force according to the target motor torque of each leg in the previous control cycle, and the ground contact force in the first control cycle is zero;

[0006] Use the target expected leg length data, actual leg length data and other control data of each leg to perform torque processing to obtain the target motor torque of each leg;

[0007] Control each leg according to the target motor torque of each leg.

[0008] Further, in one embodiment, the admittance control is performed based on the initial expected leg length data and the ground contact force of each leg to obtain the target expected leg length data of each leg, including: obtaining first processing information of each leg based on the initial expected leg length data of each leg; wherein, the first processing information includes the change acceleration and change rate of the initial expected leg length data; and obtaining the target expected leg length data of each leg according to the first processing information and the ground contact force of each leg in combination with a preset admittance control equation.

[0009] Further, in one embodiment, the target motor torque includes the knee joint motor torque and other target motor torques; the process of using the target expected leg length data, the actual leg length data and other control data of each leg to obtain the target motor torque of each leg includes: using the target expected leg length data and the actual leg length data of each leg to perform torque processing to obtain the knee joint motor torque of each leg; and using the other control data of each leg to perform torque processing to obtain the other target motor torques of each leg.

[0010] Further, in one embodiment, the process of using the target expected leg length data and the actual leg length data of each leg to perform torque processing to obtain the knee joint motor torque of each leg includes: obtaining second processing information based on the actual leg length data and the target expected leg length data of each leg; wherein, the second processing information includes the change acceleration and change rate of the target expected leg length data and the change rate of the actual leg length data; obtaining the target output force of each leg according to the second processing information, the actual leg length data and the target expected leg length data of each leg; and obtaining the knee joint motor torque of each leg according to the target output force of each leg.

[0011] Further, in one embodiment, the process of estimating the contact force according to the target motor torque of each leg in the previous control cycle includes: obtaining an equation of the joint space acceleration of each leg with respect to the ground contact force in combination with the generalized coordinates of the two-wheeled foot robot according to the target motor torque of each leg in the previous control cycle; performing wheel rolling constraint processing on each leg to obtain the task space acceleration of the contact point of each leg with the ground; and obtaining the ground contact force according to the equation of the joint space acceleration of each leg with respect to the ground contact force and the task space acceleration of the contact point of each leg with the ground.

[0012] Further, in one embodiment, obtaining the equation of the joint space acceleration of each leg with respect to the ground contact force according to the target motor torque of each leg in the previous control cycle and combining the generalized coordinates of the bipedal robot includes: obtaining the dynamic model of the bipedal robot according to the target motor torque of each leg in the previous control cycle and combining the generalized coordinates of the bipedal robot; performing an inverse operation on the mass matrix in the dynamic model to obtain the equation of the joint space acceleration of each leg with respect to the ground contact force.

[0013] Further, in one embodiment, performing a wheel rolling constraint process on each leg to obtain the task space acceleration of the contact point of each leg with the ground includes: obtaining the task space velocity of the contact point of each leg with the ground based on the generalized coordinates of the bipedal robot; performing a derivative process on the task space velocity of the contact point of each leg with the ground to obtain the task space acceleration of the contact point of each leg with the ground.

[0014] Further, in one embodiment, obtaining the ground contact force according to the equation of the joint space acceleration of each leg with respect to the ground contact force and the task space acceleration of the contact point of each leg with the ground includes: determining the task space acceleration of the contact point of each leg with the ground to be zero to obtain the ground rolling constraint state equation of each leg; processing the ground rolling constraint state equation of each leg to obtain the joint space acceleration of each leg; obtaining the ground contact force according to the joint space acceleration of each leg and the equation of the joint space acceleration of each leg with respect to the ground contact force.

[0015] On the other hand, an admittance control device for a bipedal robot provided in an embodiment of the present application includes:

[0016] An admittance controller, configured to perform admittance control based on the initial expected leg length data of each leg and the ground contact force to obtain the target expected leg length data of each leg; wherein, the ground contact force is obtained by estimating the contact force according to the target motor torque of each leg in the previous control cycle, and the ground contact force in the first control cycle is zero;

[0017] A torque processor, configured to perform torque processing using the target expected leg length data, the initial expected leg length data, and other control data of each leg to obtain the target motor torque of each leg;

[0018] A leg controller, configured to independently control each leg according to the target motor torque of each leg.

[0019] In yet another aspect, an embodiment of the present application provides a bipedal robot, including:

[0020] At least one processor;

[0021] At least one memory for storing at least one program;

[0022] When the at least one program is executed by the at least one processor, the at least one processor implements the admittance control method of the above-mentioned two-wheeled biped robot.

[0023] The beneficial effects of this application are as follows: A method, device and two-wheeled biped robot for admittance control of a two-wheeled biped robot are provided. First, admittance control is performed based on the initial expected leg length data of each leg and the ground contact force to obtain the target expected leg length data of each leg; wherein, the ground contact force is obtained by estimating the contact force according to the target motor torque of each leg in the previous control cycle, and the ground contact force in the first control cycle is zero; then, torque processing is performed using the target expected leg length data, actual leg length data and other control data of each leg to obtain the target motor torque of each leg; finally, each leg is controlled according to the target motor torque of each leg. This application can enable the two-wheeled biped robot to adapt to various complex terrains and maintain the height stability of its head, thereby effectively improving the control accuracy of the two-wheeled biped robot.

[0024] Other features and advantages of this application will be described in the subsequent specification, and part of them will be obvious from the specification, or will be understood by implementing this application. The objectives and other advantages of this application can be achieved and obtained through the structures specifically pointed out in the specification, claims and drawings. Brief Description of the Drawings

[0025] Figure 1 is a flowchart of a method for admittance control of a two-wheeled biped robot provided by this application;

[0026] Figure 2 is a control schematic diagram of a two-wheeled biped robot provided by this application;

[0027] Figure 3 is an example diagram of a simplified model provided by this application;

[0028] Figure 4 is an example diagram of a complete robot modeling provided by this application;

[0029] Figure 5 is an example diagram of a ground contact model of a two-wheeled biped robot provided by this application;

[0030] Figure 6 is a structural diagram of an admittance control device for a two-wheeled biped robot provided by this application;

[0031] Figure 7It is an example diagram of a two-wheeled and legged robot provided by this application. Detailed implementation manners

[0032] In order to make the objectives, technical solutions and advantages of this application more clear and understandable, the following further elaborates on this application in combination with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not used to limit this application.

[0033] The following further explains this application in combination with the specification drawings and specific embodiments. The described embodiments should not be regarded as limitations to this application. All other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of this application.

[0034] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments. However, it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments and can be combined with each other without conflict.

[0035] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0036] With the rapid development of robot technology, two-wheeled and legged robots have received extensive attention due to their high adaptability and flexibility in complex terrains. When a robot executes tasks, it may need to face various complex road surfaces. The two-wheeled and legged robot combines the speed advantage of a wheeled robot and the maneuvering and obstacle-crossing ability of a legged robot, and can move efficiently and flexibly under various complex terrains, having broad application prospects.

[0037] The Swiss Federal Institute of Technology Zurich modified the quadruped robot ANYmal, changed the point-foot robot into a motor-driven four-wheeled and legged robot, and obtained the nonholonomic constraints of the wheel-foot contact points by modeling the ground contact rolling constraints, and finally achieved whole-body motion control. Subsequently, the offline trajectory planning of the Zero Moment Point (ZMP) and the upper-level planning of the Model Predictive Control (MPC) were realized.

[0038] For the control method of a two-wheeled biped robot, it is usually achieved by building a wheeled inverted pendulum model (WIP) and using the state feedback of a linear quadratic regulator (LQR). In addition, there is another method which is to use a whole-body motion control algorithm. Through a further simplified rolling contact model, the non-holonomic constraints of the contact point are written as a task space control mode, and through a simplified model of an inverted pendulum in the sagittal plane, the overall moment of inertia of the robot is used as a control task, with the state feedback of the linear quadratic regulator as the tracking target, realizing a control framework that combines the linear quadratic regulator and whole-body motion control, which can achieve flexible control and a motion mode that combines rolling and walking, and it can avoid obstacles by raising one foot end.

[0039] Accordingly, at present, most related technologies use a planar inverted pendulum model to model the two-wheeled biped robot, which simplifies the legs of the two-wheeled biped robot into a link with variable length for calculation. However, this method ignores the kinematic and dynamic characteristics of the robot's legs. Moreover, this method more often models the two-wheeled biped robot as a robot with a fixed base, ignoring the floating base nature of the two-wheeled biped robot. The above reasons will lead to a reduction in the modeling accuracy of the two-wheeled biped robot, resulting in limited motion performance of the two-wheeled biped robot.

[0040] In addition, due to the lack of control in the depth axis direction of the current two-wheeled biped robot, the current two-wheeled biped robot usually ensures that it will not roll over during movement by fixing the desired leg length and the roll angle direction, thereby improving its stability during movement. However, in practical applications, the leg length of the two-wheeled biped robot will affect the stability of its head. When facing complex terrains such as slopes, the two-wheeled biped robot often tends to use a proportional-integral-derivative (PID) controller to maintain a fixed desired leg length, while ignoring the flexible control of its legs, which results in the two-wheeled biped robot being difficult to maintain the stability of its head under global control, and its legs being difficult to adapt to complex terrains, thus leading to a decrease in the control accuracy of the two-wheeled biped robot.

[0041] In view of this, an admittance control method, device, and two-wheeled and legged robot of the present application embodiment are provided. First, the present application embodiment fully considers the floating base characteristics and leg structure of the two-wheeled and legged robot, models it as a full-body floating base model, improves the modeling accuracy of the two-wheeled and legged robot, and at the same time fully considers the mechanical characteristics of the two-wheeled and legged robot to improve the solution effect of the kinematic model of the two-wheeled and legged robot. Secondly, the present application embodiment estimates the ground contact force based on the above model, judges the terrain condition of the sole of the two-wheeled and legged robot through the estimated ground contact force, and performs admittance control on the legs of the two-wheeled and legged robot, so that the two-wheeled and legged robot can adaptively change the leg length based on the terrain condition, ensure that the two-wheeled and legged robot can adapt to complex terrains, and at the same time improve the stability of its head under global control, thereby effectively improving the control accuracy of the two-wheeled and legged robot.

[0042] First, a method for admittance control of a two-wheeled and legged robot provided by an embodiment of the present application will be described in detail below with reference to the accompanying drawings.

[0043] A method for admittance control of a two-wheeled and legged robot provided by an embodiment of the present application can be applied to a terminal, can also be applied to a server, or can also be software running on a terminal or a server, etc. The terminal can be a tablet computer, a notebook computer, a desktop computer, etc., but is not limited thereto. The server can be an independent physical server, can also be a server cluster or a distributed system composed of multiple physical servers, can also be a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, Content Delivery Network (CDN), and big data and artificial intelligence platforms. In addition, the server can also be a node server in a blockchain network, but is not limited thereto. Among them, the blockchain is a new application mode of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, and encryption algorithms.

[0044] Refer to Figure 1 , Figure 1 is a flowchart of a method for admittance control of a two-wheeled and legged robot provided by the present application. The two-wheeled and legged robot is usually configured with two legs, namely a left leg and a right leg. The admittance control method can include the following steps S101-S103.

[0045] S101, perform admittance control based on the initial expected leg length data and ground contact force of each leg to obtain the target expected leg length data of each leg.

[0046] It should be noted that the initial expected leg length data refers to the initially expected leg length, while the target expected leg length data refers to the leg length after being adjusted by admittance control. In addition, the ground contact force refers to the contact force between the wheels of each leg and the ground, which is obtained by estimating the contact force based on the target motor torque of each leg in the previous control cycle.

[0047] It can be understood that the ground contact force in the first control cycle is defined as zero, which means that the first control cycle will not affect the target expected leg length data, equivalent to only the first control cycle without the participation of admittance control, and the ground contact force in all subsequent control cycles will be estimated based on the target motor torque of the previous control cycle.

[0048] In this step, according to prior knowledge, when the bipedal wheeled robot encounters an uphill terrain, the contact force between the bipedal wheeled robot and the ground will have an upward impact trend; when the bipedal wheeled robot encounters a downhill terrain, the contact force between the bipedal wheeled robot and the ground will have a downward trend. Accordingly, in this step, the leg length of the bipedal wheeled robot is dynamically adjusted through the change of the contact force to make it adapt to complex terrains and ensure stable control of the height of the head. Specifically, in this step, the admittance control method is adopted to control the leg lengths of the left and right legs of the robot, that is, based on the initial expected leg length data and the ground contact force of each leg, admittance control is performed, aiming to correct the initially expected leg length through admittance control, so as to obtain the target expected leg length data of each leg.

[0049] S102, Use the target expected leg length data, actual leg length data and other control data of each leg to perform torque processing to obtain the target motor torque of each leg.

[0050] It should be noted that the target motor torque refers to the torque of the motors of the joints and wheels of the legs. In addition, the other control data refers to the data associated with controllers such as the balance controller, roll angle controller, and yaw angle controller of the bipedal wheeled robot.

[0051] In this step, after obtaining the target expected leg length data, first, obtain the actual leg length data of the bipedal wheeled robot; then, based on using the target expected leg length data, actual leg length data and other control data of each leg to perform torque processing, aiming to determine the target motor torque of each leg, and the target motor torque will be used to control the joints and wheels of each leg.

[0052] S103, Control each leg according to the target motor torque of each leg.

[0053] In this step, the left leg of the two-wheeled foot robot is controlled based on the target motor torque of the left leg so that the joint motor and the wheel motor of the left leg reach the corresponding motor torque, and the right leg of the two-wheeled foot robot is controlled based on the target motor torque of the right leg so that the joint motor and the wheel motor of the right leg reach the corresponding motor torque, thereby realizing the control processing of the two-wheeled foot robot.

[0054] It can be seen that for the control method of the two-wheeled foot robot, on the one hand, the embodiment of the present application does not need to rely on external sensing devices such as laser radar, force sensor, etc., and introduces the method of estimating ground contact force without external sensing, which can enable the two-wheeled foot robot to perceive the terrain conditions of the sole of the foot more quickly and accurately, so that the two-wheeled foot robot can adaptively change the leg length based on the terrain conditions; on the other hand, the embodiment of the present application realizes flexible control of the legs of the two-wheeled foot robot through admittance control, which can enable the two-wheeled foot robot to exhibit dynamic characteristics similar to a spring-damper system when interacting with complex terrain. This characteristic enables the two-wheeled foot robot to freely adjust its own motion state on surfaces of different hardness, so that the actual leg length of the two-wheeled foot robot is closer to the expected leg length; the embodiment of the present application can effectively ensure that the two-wheeled foot robot can adapt to complex terrain, while improving the stability of its head under global control, thereby effectively improving the control accuracy of the two-wheeled foot robot.

[0055] The above steps will be further described below.

[0056] In some embodiments, reference Figure 2 , Figure 2 is a control principle diagram of the two-wheeled leg robot provided by the present application; in the above step S101, the admittance control is performed based on the initial expected leg length data of each leg and the ground contact force, and the implementation process of obtaining the target expected leg length data of each leg may include the following steps S201-S202:

[0057] S201, based on the initial expected leg length data of each leg, obtaining first processing information of each leg; wherein the first processing information includes the change acceleration and change rate of the initial expected leg length data;

[0058] S202, obtaining target expected leg length data of each leg according to the first processed information and the ground contact force of each leg in combination with a preset admittance control equation.

[0059] In this embodiment, in the traditional control algorithm of the two-wheeled leg robot, there is rarely a way to flexibly control the legs of the two-wheeled leg robot, which makes it difficult for the two-wheeled leg robot to adapt to complex terrain and cannot ensure the stability of its head during movement, thereby causing the control accuracy of the two-wheeled leg robot to decrease. In this regard, the following admittance control process is performed for both the left and right legs of the two-wheeled leg robot:

[0060] First, take the derivative of the initial expected leg length data of the leg to obtain the first derivative of the initial expected leg length data of the leg as the change rate of the initial expected leg length data of the leg, and obtain the second derivative of the initial expected leg length data of the leg as the change acceleration of the initial expected leg length data of the leg, thus obtaining the first processing information of the leg.

[0061] Then, construct the admittance control equation shown in formula (1) as follows:

[0062]

[0063] In formula (1), L′ leg，d represents the adjusted expected leg length of the leg, that is, the target expected leg length data; represents the change rate of the target expected leg length data of the leg; represents the change acceleration of the target expected leg length data of the leg; L leg，d represents the original input leg length of the leg, that is, the initial expected leg length data; represents the change rate of the initial expected leg length data of the leg; represents the change acceleration of the initial expected leg length data of the leg; K represents the stiffness coefficient, which determines the degree of change of the leg when subjected to an external force; B represents the damping coefficient, which is related to the change speed of the leg length and affects the absorption and shock absorption effect of the leg on the external force; M represents the inertia coefficient, which is related to the change acceleration of the leg length and affects the acceleration response of the leg to the external force; F represents the ground contact force. It can be understood that the change rate of the leg length data is the first derivative of the leg length data, and the change acceleration of the leg length data is the second derivative of the leg length data.

[0064] Since it finally needs to be implemented in the robot main control code, in order to ensure the accuracy of leg control, discretization processing is required. In this embodiment, the forward difference equation is used to discretize the admittance control equation shown in formula (1) above, that is, discretize the change rate and change acceleration of the initial expected leg length data and substitute them into the admittance control equation. The change rate and change acceleration of the target expected leg length data can also be discretized and substituted into the admittance control equation in the same way. Taking the initial expected leg length data as an example, its discretization process can be expressed as formula (2) below:

[0065]

[0066] In formula (2), T represents the sampling time, that is, the time interval between two adjacent sampling periods; k represents the sampling period, which can be understood as the control period.

[0067] Next, substitute the discretized data into the admittance control equation shown in the above formula (1), and perform Laplace transformation on the admittance control equation to obtain the admittance control equation after Laplace transformation, as shown in the following formula (3):

[0068] (L′ leg,d -L leg,d )K+(L′ leg,d -L leg,d )Bs+(L′ leg,d -L leg,d )Ms 2 =k ad F (3);

[0069] In formula (3), s represents the Laplace operator; k ad represents the compensation coefficient, which is a parameter set to appropriately adjust the compensation ratio of the robot's ground contact force and leg length.

[0070] Finally, by performing parameter transformation on the above admittance control equation after Laplace transformation, the target expected leg length data shown in the following formula (4) can be obtained:

[0071]

[0072] Optionally, the damping coefficient, inertia coefficient, and stiffness coefficient in the admittance control equation can be calibrated according to the actual situation, and this embodiment does not make specific limitations in this regard.

[0073] It should be noted that when going uphill and downhill, the external force change modes at the contact points between the wheels of the two-wheeled foot robot and the ground are different. When going uphill, on the basis of the original ground force remaining unchanged, there is another external force perpendicular to the slope direction generated by the collision between the wheels and the slope; when going downhill, the two-wheeled foot robot often converts the upward supporting force into an external force perpendicular to the slope, and the peak values of these two force changes are different. In addition, affected by gravity, the two-wheeled foot robot will have a faster speed when going downhill, so the two-wheeled foot robot also requires a faster reaction speed when going downhill. Since the external force changes of the two-wheeled foot robot are different and the required reaction times are also different, if the same set of admittance control coefficients is used to control the two-wheeled foot robot when going uphill and downhill, it will lead to poor actual performance of the two-wheeled foot robot, such as insufficient reaction speed and excessive leg retraction when going downhill. Therefore, this embodiment uses different admittance control coefficients for targeted control, so as to enable the head of the two-wheeled foot robot to adapt to the uphill and downhill states simultaneously.

[0074] Specifically, in the admittance control equation, the damping coefficient, the inertia coefficient, and the stiffness coefficient together constitute the admittance control coefficient of the admittance control equation, where: for the stiffness coefficient K, the stiffness coefficient of the two-wheeled and two-legged robot when going downhill needs to be greater than that when going uphill; for the damping coefficient B, the damping coefficient of the two-wheeled and two-legged robot when going downhill needs to be less than that when going uphill; for the inertia coefficient M, there is no requirement for a specific magnitude relationship. The inertia coefficient of the two-wheeled and two-legged robot when going downhill can be equal to that when going uphill. Of course, the two can also be unequal, but it should be noted that the difference between the inertia coefficients when going uphill and downhill should not be too large, that is, the absolute value of the difference between the inertia coefficient of the two-wheeled and two-legged robot when going downhill and the inertia coefficient when going uphill needs to be less than a preset threshold. In some cases, the inertia coefficient of the two-wheeled and two-legged robot when going downhill can be slightly less than that when going uphill. In this way, by introducing an adaptive parameter mechanism and adopting different admittance control coefficients for different terrains, the two-wheeled and two-legged robot can have better adaptability when facing different terrains, improving the adaptability of the control algorithm and the motion stability of the two-wheeled and two-legged robot.

[0075] For example, the inertia coefficient of the two-wheeled and two-legged robot when going uphill can be 1, its stiffness coefficient can be 0.2, and its damping coefficient can be 5; the inertia coefficient of the two-wheeled and two-legged robot when going downhill can be 0.8, its stiffness coefficient can be 10, and its damping coefficient can be 3, but it is not limited to this.

[0076] Optionally, the compensation coefficient can be calibrated according to the actual situation, and this embodiment does not make specific limitations on this.

[0077] It should be noted that the requirements for the control dynamic performance of the two-wheeled and two-legged robot when going uphill and downhill are different. For example, when going downhill, it is often required that the response speed of the two-wheeled and two-legged robot be faster to adapt to the faster downhill speed of the two-wheeled and two-legged robot, prompting the two-wheeled and two-legged robot to recover to the target desired leg length data faster. In addition, the peak values of the ground contact forces received by the two-wheeled and two-legged robot when going uphill and downhill are different. Accordingly, the compensation coefficient of the two-wheeled and two-legged robot when going downhill needs to be greater than that when going uphill so that the two-wheeled and two-legged robot can compensate for the peak difference when going downhill.

[0078] It can be seen that in this embodiment, the admittance control method is adopted to flexibly control the legs of the two-wheeled biped robot, so that it can better adapt to complex terrains and maintain the height stability of the head. Through admittance control, the two-wheeled biped robot can exhibit dynamic characteristics similar to a spring-damper system when interacting with complex terrains. This characteristic enables the two-wheeled biped robot to freely adjust its motion state on surfaces with different hardnesses, making the actual leg length of the two-wheeled biped robot more tend to the desired leg length, better adapting to complex terrains, and enhancing the stability of its head under global control, thereby effectively improving the control accuracy of the two-wheeled biped robot.

[0079] In some embodiments, referring to Figure 2 , the above-mentioned target motor torques may include knee joint motor torques and other target motor torques; in the above step S102, the process of obtaining the target motor torques of each leg by performing torque processing using the target desired leg length data, actual leg length data, and other control data of each leg may include the following steps S301-S302:

[0080] S301, performing torque processing using the target desired leg length data and actual leg length data of each leg to obtain the knee joint motor torques of each leg;

[0081] S302, performing torque processing using the other control data of each leg to obtain the other target motor torques of each leg.

[0082] It should be noted that the other target motor torques may include, but are not limited to, hip joint motor torques, inner motor torques, outer motor torques, and wheel motor torques.

[0083] In this embodiment, the controller of the two-wheeled biped robot is relatively complex. Usually, different controllers are required to calculate the torques of motors in different parts, including a balance controller, a height controller, a roll angle controller, a yaw angle controller, etc. Among them, the balance controller affects the torques of the wheel motor and the hip joint motor, the height controller affects the knee joint motor torque, the roll angle controller compensates the knee joint motor torque, and the yaw angle controller compensates the wheel motor torque. Finally, the two-wheeled biped robot is controlled based on all these torques. Accordingly, in this embodiment, the following torque processing process is performed on both the left and right legs of the two-wheeled biped robot: performing torque processing using the target desired leg length data and actual leg length data of the leg in combination with the above-mentioned height controller to obtain the knee joint motor torque of the leg, and at the same time performing torque processing using the other control data of the leg in combination with the other controllers except the height controller to obtain the other joint torques except the knee joint motor torque, such as hip joint motor torques, inner motor torques, outer motor torques, and wheel motor torques, etc., and then obtaining the target motor torque of the leg, so as to effectively ensure the control accuracy of the two-wheeled biped robot.

[0084] It should be noted that in the embodiments of the present application, only the height controller is improved, while other controllers are all prior arts. The specific content thereof (i.e., the specific control methods of other controllers and other control data) can be directly cited from the technology in the paper "DIABLO: A 6-DoF Wheeled Bipedal Robot Composed Entirely of Direct-Drive Joints".

[0085] In some embodiments, referring to Figure 2 , in the above step S301, the process of obtaining the knee joint motor torque of each leg by performing torque processing on the target desired leg length data and the actual leg length data of each leg may include the following steps S401 - S403:

[0086] S401, based on the actual leg length data and the target desired leg length data of each leg, obtain second processing information; wherein, the second processing information includes the change acceleration and change rate of the target desired leg length data and the change rate of the actual leg length data;

[0087] S402, according to the second processing information, the actual leg length data and the target desired leg length data of each leg, obtain the target output force of each leg;

[0088] S403, according to the target output force of each leg, obtain the knee joint motor torque of each leg.

[0089] It should be noted that the target output force refers to the output force combined by a Proportional-Derivative (PD) controller and a feedforward controller, which can be understood as the force required to change the legs of the bipedal wheeled robot without considering the head mass, and this force is used to adjust the height change according to the input.

[0090] In this embodiment, for the height controller of the bipedal wheeled robot, its input is the target desired leg length data of each leg, and its output is the knee joint motor torque of each leg. Specifically, for the left leg and the right leg of the bipedal wheeled robot, the process of solving the knee joint motor torque is as follows:

[0091] First, take the derivative of the actual leg length data of the leg to obtain the first derivative of the actual leg length data of the leg as the change rate of the actual leg length data of the leg, and, take the derivative of the target desired leg length data of the leg to obtain the first derivative of the target desired leg length data of the leg as the change rate of the target desired leg length data of the leg, and obtain the second derivative of the target desired leg length data of the leg as the change acceleration of the target desired leg length data of the leg, thereby obtaining the second processing information.

[0092] Then, based on the second processing information of the leg, the actual leg length data, and the target desired leg length data, and in combination with the preset proportional derivative controller coefficients, the target output force of the leg is obtained, which can be expressed by the following formula (5):

[0093]

[0094] In formula (5), ΔF represents the target output force; L′ leg，d represents the target desired leg length data of the leg; represents the change rate of the target desired leg length data of the leg; represents the change acceleration of the target desired leg length data of the leg; L leg represents the actual leg length data of the leg; represents the change rate of the actual leg length data of the leg; k p and k d are both parameters of the proportional derivative controller, which are preset values.

[0095] Finally, based on the target output force of the leg, in combination with the head mass and the head Jacobian matrix of the bipedal robot, the knee joint motor torque of the leg is obtained, which can be expressed by the following formula (6), where the head Jacobian matrix refers to the Jacobian matrix that maps the head lifting force of the bipedal robot to the knee joint motor torque:

[0096] τ knee =J T (ΔF+m H g) (6);

[0097] In formula (6), τ knee represents the knee joint motor torque of the leg; J T represents the head Jacobian matrix of the bipedal robot; m H represents the head mass of the bipedal robot; g represents the acceleration due to gravity.

[0098] It can be seen that in this embodiment, the target desired leg length data obtained by admittance control is used as the benchmark for the head height control of the bipedal robot, and in combination with the height controller, the acquisition process of the knee joint motor torque of the leg is realized. In this way, the accuracy of the knee joint motor torque of the leg can be improved, enabling the bipedal robot to better adapt to complex terrains and enhancing the stability of its head under global control, thereby effectively improving the control accuracy of the bipedal robot.

[0099] In some embodiments, for the above-mentioned ground contact force, the implementation process of estimating the contact force according to the target motor torques of each leg in the previous control cycle may include the following steps S501 - S503:

[0100] S501, according to the target motor torque of each leg in the previous control cycle, combined with the generalized coordinates of the two-wheeled foot robot, obtain the equation of the joint space acceleration of each leg with respect to the ground contact force;

[0101] S502, performing wheel rolling constraint processing on each leg to obtain the task space acceleration of the contact point between each leg and the ground;

[0102] S503, obtaining the ground contact force according to the equation of the joint space acceleration of each leg with respect to the ground contact force and the task space acceleration of the contact point between each leg and the ground.

[0103] It should be noted that the task space acceleration of the contact point between the leg and the ground refers to the acceleration of the end effector of the two-wheeled leg robot moving in the task space.

[0104] In this embodiment, the force exerted by the left and right wheels of the two-wheeled leg robot in contact with the ground is estimated by the current state of the two-wheeled leg robot, so as to indirectly obtain the current terrain state, thereby facilitating dynamic control of the leg length of the two-wheeled leg robot and making the head of the two-wheeled leg robot more stable. Specifically,

[0105] In terms of ground contact force estimation, for the current control cycle, the following ground contact force estimation process is performed for both the left and right legs of the two-wheeled foot robot:

[0106] Firstly, considering the floating basis characteristics, kinematic characteristics and dynamic characteristics of the two-wheeled leg robot, based on the target motor torque of the leg in the previous control cycle and combined with the generalized coordinates of the two-wheeled leg robot (which includes the joint coordinates of the two-wheeled leg robot and the undriven base coordinates in the inertial coordinate system), an equation for the joint space acceleration of the leg with respect to the ground contact force is constructed. This equation is intended to be combined with the subsequent rolling constraint to calculate the contact force between the leg and the ground.

[0107] At the same time, in order to limit the movement of the wheels on the ground, it is necessary to establish the constraint equation of wheel rolling and derive the corresponding constraint force. Based on this, the wheel rolling constraint processing of the leg is carried out, the constrained rolling of the leg wheel is constructed, and the task space acceleration of the contact point between the leg and the ground, that is, the speed constraint, is calculated based on the constrained rolling.

[0108] Then, the equation of the joint space acceleration of the leg with respect to the ground contact force and the task space acceleration of the contact point between the leg and the ground are combined to obtain the contact force between the leg and the ground, that is, the ground contact force.

[0109] It can be seen that the present embodiment does not need to rely on external sensing devices such as lidar, force sensors, etc., and can accurately estimate the ground contact force without external sensing based on the dynamic model of the two-wheeled leg robot itself, thereby enabling the two-wheeled leg robot to sense the terrain conditions of the soles of the feet more quickly and accurately, so that the two-wheeled leg robot can adaptively change the length of its legs based on the terrain conditions and maintain the high stability of its head, thereby improving the control accuracy of the two-wheeled leg robot.

[0110] In some embodiments, reference Figure 3 and Figure 4 , Figure 3 is an example diagram of a simplified model (one side of the original robot configuration) provided by this application. Figure 4 is an example diagram of complete modeling of the robot provided by the present application; in the above step S501, according to the target motor torque of each leg in the previous control cycle, combined with the generalized coordinates of the two-wheeled foot robot, the implementation process of obtaining the equation of the joint space acceleration of each leg with respect to the ground contact force can include the following steps S601-S602:

[0111] S601, obtaining a dynamic model of the two-wheeled leg robot according to the target motor torque of each leg in the previous control cycle and the generalized coordinates of the two-wheeled leg robot;

[0112] S602, performing inversion processing on the mass matrix in the dynamic model to obtain an equation of the joint space acceleration of each leg with respect to the ground contact force.

[0113] In this embodiment, since the two-wheeled foot robot is a robot with a floating base, its floating base characteristics need to be fully considered, so the base of the two-wheeled foot robot needs to be regarded as connected to a six-degree-of-freedom joint. Define the inertial coordinate system I, the floating base coordinate system B and the wheel coordinate system C. The wheel coordinate system includes the left wheel coordinate system C l and the right wheel coordinate system C r , the floating base coordinate system B is located in the vertical direction of the midpoint of the line connecting the two wheel contact points. The midpoint of the line connecting the two wheel contact points is defined as point N, and the floating base is defined as point B, such as Figure 3 As shown, the red arrow represents the horizontal axis of the coordinate system, the green arrow represents the vertical axis of the coordinate system, and the blue arrow represents the depth axis of the coordinate system. All coordinate systems have their depth axes (Z) oriented along the ground normal vector, their horizontal axes (X) oriented along the forward direction of the two-wheeled robot, and their vertical axes (Y) defined along the line connecting the contact points of the two wheels.

[0114] First, a kinematic tree model of the two-wheeled foot robot is built, and the joints and rigid body rods of the two-wheeled foot robot are numbered and encoded. Since the legs of the two-wheeled foot robot are closed-loop legs, and the existing processing methods usually require additional processing of the closed-loop legs, such as disconnecting passive joints and applying closed-loop constraints. This method is often very complex and not necessary for the control of the two-wheeled foot robot. Therefore, in this embodiment, the legs of the two-wheeled foot robot are simplified to open-loop legs, as Figure 3 and Figure 4 shown. By simplifying the model without affecting the experiment, the subsequent efficiency can be effectively improved.

[0115] It should be noted that in Figure 3 and Figure 4 , B is the rigid body component of the two-wheeled foot robot, J is the connecting joint between the rigid body components of the two-wheeled foot robot, and the parameter definitions in the figure are shown in Table 1 below.

[0116] Table 1: Parameter Definitions

[0117]

[0118]

[0119] Next, according to the above content, the generalized coordinates of the kinematic generation tree model of the two-wheeled foot robot are defined. The generalized coordinates include the joint coordinates of the two-wheeled foot robot in the inertial coordinate system and the undriven base coordinates; as shown in the following formula (7):

[0120]

[0121] In formula (7), q represents the generalized coordinates of the two-wheeled foot robot, q b ∈R 3 ×SO represents the undriven base coordinates, q j ∈R 6 represents the joint coordinates; in q b In, I r IB represents the position vector of point B relative to the origin of the inertial coordinate system I and is represented in the inertial coordinate system I. The above annotation usage applies to the subsequent embodiments of this application, r IB represents the translation of the base of the two-wheeled foot robot (positive if the translation is in the positive direction of the coordinate axis, otherwise negative); R IB ∈SO represents the rotation matrix of the two-wheeled foot robot; in q j, q1 to q6 represent the generalized coordinates of the left medial motor / hip motor, the left lateral motor / knee motor, the left wheel motor, the right medial motor / hip motor, the right lateral motor / knee motor, and the right wheel motor, respectively.

[0122] At the same time, the six motor torques of the two-wheeled foot robot are defined as shown in the following formula (8):

[0123]

[0124] In formula (8), τ represents the torque term; τ a ∈R 6 represents the torque of the driven joint, where τ1 to τ6 represent the left medial motor / hip joint motor torque, the left lateral motor / knee joint motor torque, the left wheel motor torque, the right medial motor / hip joint motor torque, the right lateral motor / knee joint motor torque and the right wheel motor torque, respectively.

[0125] Afterwards, for the left and right legs of the two-wheeled foot robot, we have: Based on the target motor torque of the legs in the previous control cycle and the generalized coordinates of the two-wheeled foot robot, the dynamic model of the two-wheeled foot robot is constructed, as shown in the following formula (9):

[0126]

[0127] In formula (9), H∈R 12×12 Represents the mass (inertia) matrix; C∈R 12 represents the Coriolis term; g represents the gravitational acceleration; S∈R 12×6 represents the selection matrix; τ represents the moment term; represents the common Jacobian matrix of the two-wheel contact points; F represents the ground contact force; Represents the robot joint space acceleration, which is the second-order derivative of the generalized coordinates of the two-wheeled foot robot. Among them, the mass matrix, Coriolis term, selection matrix, torque term and Jacobian matrix common to the two-wheeled foot contact point can all be obtained by calculation or the body of the two-wheeled foot robot. Since the subsequent admittance control is for the control of the ground contact force, the construction of the dynamic model here is very critical.

[0128] Finally, by inverting the mass matrix in the dynamic model shown in the above formula (9), we can obtain the equation of the joint space acceleration of the leg with respect to the ground contact force shown in the following formula (10). This equation is applicable to a single leg, that is, the left leg or the right leg. This equation aims to calculate the magnitude of the ground contact force of the left and right ends of the robot by listing the equation of the joint space acceleration of the robot with respect to the ground contact force and combining it with the subsequent rolling constraint:

[0129]

[0130] It can be seen that in this embodiment, considering the floating base characteristics of the bipedal wheeled robot and the leg structure, the bipedal wheeled robot is modeled as a full-body floating base model, which can effectively improve the modeling accuracy of the bipedal wheeled robot; at the same time, this embodiment also fully considers the mechanical characteristics of the bipedal wheeled robot, so as to effectively improve the solution effect of the kinematic model of the bipedal wheeled robot.

[0131] In some embodiments, referring to Figure 5 , Figure 5 is an example diagram of the ground contact model of the bipedal wheeled robot provided by this application; in the above step S502, the implementation process of obtaining the task space acceleration of the contact points between each leg and the ground by performing wheel rolling constraint processing on each leg may include the following steps S701 - S702:

[0132] S701, based on the generalized coordinates of the bipedal wheeled robot, obtain the task space velocity of the contact points between each leg and the ground;

[0133] S702, perform a derivative processing on the task space velocity of the contact points between each leg and the ground to obtain the task space acceleration of the contact points between each leg and the ground.

[0134] In this embodiment, in order to restrict the movement of the wheels on the ground, it is necessary to establish a constraint equation for wheel rolling and derive the corresponding binding force. The contact points between the left and right wheels of the bipedal wheeled robot and the ground (i.e., point C l and point C r ) are parameterized relative to the wheel coordinate system C according to Contour Kinematics combined with ContourParameter, and the ContourParameter depends on the contact between the wheel and the ground. As shown in Figure 5 , W represents the position of the wheel center. The coordinates of the contact points between the left and right wheels and the ground are determined by the ContourParameter σ. In this embodiment, it is assumed that the contact points between the left and right wheels of the robot and the ground are always directly below the wheel center position W. Therefore, the ground normal vector I of the contact points between the left and right wheels and the ground nAlways vertically upward. Under this condition, in this embodiment, the Jacobian matrix that maps the joint space velocity of the leg to the task space velocity of the contact point between the leg and the ground is obtained. This Jacobian matrix can be pre-calibrated. Since the generalized coordinates of the bipedal robot are known quantities, the initial velocity constraint can be directly calculated, that is, the task space velocity of the contact point between the leg and the ground (which refers to the velocity at which the end effector of the bipedal robot moves in the task space and describes how fast and in what direction the position of the end effector changes in the task space it is in). It can be defined in the form of the Jacobian matrix as shown in the following formula (11):

[0135]

[0136] In formula (11), represents the task space velocity of the contact point between the leg and the ground; represents the first derivative of the generalized coordinates of the bipedal robot, that is, the joint space velocity; J C represents the Jacobian matrix that maps the joint space velocity of the leg to the task space velocity of the contact point between the leg and the ground.

[0137] Finally, by taking the derivative of the task space velocity of the contact point between the leg and the ground shown in the above formula (11), the task space acceleration of the contact point between the leg and the ground shown in the following formula (12) can be obtained, which is the final velocity constraint:

[0138]

[0139] In formula (12), represents the task space acceleration of the contact point between the leg and the ground, which is the first derivative of the task space velocity of the contact point between the leg and the ground; represents the first derivative of the Jacobian matrix that maps the robot joint velocity to the task space velocity of the contact point between the leg and the ground.

[0140] In some embodiments, in the above step S503, the process of obtaining the ground contact force according to the equation of the joint space acceleration of each leg with respect to the ground contact force and the task space acceleration of the contact point between each leg and the ground may include the following steps S801 - S803:

[0141] S801, determine the task space acceleration of the contact point between each leg and the ground to be zero, and obtain the ground rolling constraint state equation of each leg;

[0142] S802, process the ground rolling constraint state equation of each leg to obtain the joint space acceleration of each leg;

[0143] S803. Obtain the ground contact force according to the joint space acceleration of each leg and the equation of the joint space acceleration of each leg with respect to the ground contact force.

[0144] In this embodiment, ground force estimation based on a dynamic model is used, so that the bipedal wheeled robot can indirectly sense terrain changes through ground force changes. For the left and right legs of the bipedal wheeled robot, the following estimation processes are both applicable:

[0145] First, assume that the contact between the leg wheel and the ground satisfies the pure rolling state, that is, the speed of the contact point between the leg and the ground is 0, and determine the task space acceleration of the contact point between the leg and the ground to be zero, then the ground rolling constraint state equation of the leg can be obtained, as shown in the following formula (13):

[0146]

[0147] Then, perform transformation processing on the ground rolling constraint state equation shown in the above formula (13), and the joint space acceleration shown in the following formula (14) can be obtained:

[0148]

[0149] Finally, substitute the joint space acceleration shown in the above formula (14) into the equation of the joint space acceleration of the leg with respect to the ground contact force shown in the above formula (10), and the ground contact force shown in the following formula (15) can be obtained:

[0150]

[0151] For the convenience of understanding the admittance control method of the above bipedal wheeled robot of the present application, an example of the actual application scenario of the admittance control method of the above bipedal wheeled robot of the present application is given here. Refer to Figure 2 , in this application scenario, the bipedal wheeled robot has two legs, namely the left leg and the right leg. For each leg, the specific implementation of its admittance control method is as follows:

[0152] S901. Leg length estimation:

[0153] In the current control cycle, first, the derivative of the initial desired leg length data of the leg is calculated to obtain the first derivative of the initial desired leg length data of the leg as the change rate of the initial desired leg length data of the leg, and the second derivative of the initial desired leg length data of the leg is obtained as the change acceleration of the initial desired leg length data of the leg; then, the above data is discretized. As shown in the above formula (2), the discretized data is substituted into the admittance control equation shown in the above formula (1), and the Laplace transform is performed on the admittance control equation to obtain the Laplace-transformed admittance control equation, as shown in the above formula (3); finally, parameter transformation is performed on the above Laplace-transformed admittance control equation to obtain the target desired leg length data shown in the following formula (4).

[0154] Among them, for the stiffness coefficient, the stiffness coefficient of the two-wheeled foot robot when going downhill needs to be greater than the stiffness coefficient of the two-wheeled foot robot when going uphill; for the damping coefficient, the damping coefficient of the two-wheeled foot robot when going downhill needs to be less than the damping coefficient of the two-wheeled foot robot when going uphill; for the inertia coefficient, the absolute value of the difference between the inertia coefficient of the two-wheeled foot robot when going downhill and the inertia coefficient of the two-wheeled foot robot when going uphill needs to be less than a preset threshold. In addition, the compensation coefficient of the two-wheeled foot robot when going downhill needs to be greater than the compensation coefficient of the two-wheeled foot robot when going uphill.

[0155] S902, torque control:

[0156] In the current control cycle, for the height controller:

[0157] In the current control cycle, first, the derivative of the actual leg length data of the leg is calculated to obtain the first derivative of the actual leg length data of the leg as the change rate of the actual leg length data of the leg, and, the derivative of the target desired leg length data of the leg is calculated to obtain the first derivative of the target desired leg length data of the leg as the change rate of the target desired leg length data of the leg, and the second derivative of the target desired leg length data of the leg is obtained as the change acceleration of the target desired leg length data of the leg; then, the above data is substituted into the above formula (5) to obtain the target output force of the leg; afterwards, the target output force of the leg, the head mass of the two-wheeled foot robot, and the head Jacobian matrix are substituted into the above formula (6) to obtain the knee joint motor torque of the leg.

[0158] In the current control cycle, for other controllers:

[0159] Use other control data of the legs and combine with other controllers except the height controller above to perform torque processing, obtaining other joint torques except the knee joint motor torque, such as hip joint motor torque, medial motor torque, lateral motor torque, and wheel motor torque, etc. The technology in the paper "DIABLO: A 6-DoF Wheeled Bipedal Robot Composed Entirely of Direct-Drive Joints" can be directly cited.

[0160] S903, wheeled foot control:

[0161] In the current control cycle, control the left leg of the wheeled bipedal robot based on the target motor torque of the left leg, so that the joint motors and wheel motors of the left leg reach the corresponding motor torques, and control the right leg of the wheeled bipedal robot based on the target motor torque of the right leg, so that the joint motors and wheel motors of the right leg reach the corresponding motor torques.

[0162] S904, loop termination judgment:

[0163] In the current control cycle, judge whether to end the control of the wheeled bipedal robot. For example, end the control of the wheeled bipedal robot when the task of the wheeled bipedal robot is completed, but not limited to this. If so, directly end the process; otherwise, enter step S05.

[0164] S905, contact force estimation:

[0165] Modeling process: In the current control cycle, first, define the generalized coordinates of the kinematic generation tree model of the wheeled bipedal robot. The generalized coordinates include the joint coordinates of the wheeled bipedal robot in the inertial coordinate system and the undriven base coordinates. At the same time, define the six motor torques of the wheeled bipedal robot, which can follow the target motor torque of the current control cycle; then, based on the target motor torque of the legs and the generalized coordinates of the wheeled bipedal robot in the previous control cycle, construct the dynamic model as shown in the above formula (9); finally, invert the mass matrix in the dynamic model shown in the above formula (9) to obtain the equation of the joint space acceleration of the legs with respect to the ground contact force shown in the above formula (10).

[0166] Rolling constraint: In the current control cycle, obtain the Jacobian matrix that maps the joint space velocity of the leg to the task space velocity of the contact point between the leg and the ground. This Jacobian matrix can be pre-calibrated. Since the generalized coordinates of the bipedal robot are known quantities, the initial velocity constraint, that is, the task space velocity of the contact point between the leg and the ground, can be obtained through the above formula (11). Then, take the derivative of the task space velocity of the contact point between the leg and the ground shown in the above formula (11), and the task space acceleration of the contact point between the leg and the ground shown in the above formula (12) can be obtained, which is the final velocity constraint.

[0167] Estimation of ground contact force: In the current control cycle, first, set the task space acceleration of the contact point between the leg and the ground to zero, and the ground rolling constraint state equation of the leg shown in the above formula (13) can be obtained; then, perform transformation processing on the ground rolling constraint state equation shown in the above formula (13), and the joint space acceleration shown in the above formula (14) can be obtained; finally, substitute the joint space acceleration shown in the above formula (14) into the equation of the joint space acceleration of the leg with respect to the ground contact force shown in the above formula (10), and the ground contact force shown in the above formula (15) can be obtained.

[0168] S906, Loop control:

[0169] Set the ground contact force shown in the above formula (17) as the ground contact force for the next control cycle, set the next control cycle as the current control cycle, and return to step S01 to achieve loop control.

[0170] In addition, referring to Figure 6 , this embodiment of the present application also provides an admittance control device for a bipedal robot, which includes:

[0171] An admittance controller 100, configured to perform admittance control based on the initial expected leg length data of each leg and the ground contact force to obtain the target expected leg length data of each leg; wherein, the ground contact force is obtained by estimating the contact force according to the target motor torque of each leg in the previous control cycle, and the ground contact force in the first control cycle is zero;

[0172] A torque processor 200, configured to perform torque processing by using the target expected leg length data, the initial expected leg length data, and other control data of each leg to obtain the target motor torque of each leg;

[0173] A leg controller 300, configured to independently control each leg according to the target motor torque of each leg.

[0174] The content in the above method embodiments is applicable to the device embodiments of the present invention. The functions specifically implemented by the device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those of the above method embodiments.

[0175] Finally, referring to Figure 7 , the embodiments of the present application also provide a two-wheeled foot robot, which includes:

[0176] At least one processor 400;

[0177] At least one memory 500, configured to store at least one program;

[0178] When the at least one program is executed by the at least one processor 400, the at least one processor 400 implements the admittance control method of the above two-wheeled foot robot.

[0179] The above memory 500, as a non-transitory network system, can be used to store non-transitory software programs and non-transitory computer-executable programs. In addition, the memory 500 may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some embodiments, the memory 500 may optionally include a memory 500 remotely disposed relative to the processor 400, and these remote memories 500 may be connected to the processor 400 through a network. Examples of the above networks include, but are not limited to, the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0180] The above memory 500 may be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM), etc. The memory 500 may store an operating system and other application programs. When implementing the technical solutions provided in the embodiments of the present specification through software or firmware, the relevant program codes are stored in the memory 500, and are called by the processor 400 to execute the methods of the embodiments of the present application.

[0181] The above processor 400 may be implemented in a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, etc., and is configured to execute relevant programs to implement the technical solutions provided in the embodiments of the present application.

[0182] In some embodiments, the above vehicle may further include:

[0183] An input / output interface for implementing information input and output;

[0184] A communication interface for implementing communication interaction between this device and other devices, which can achieve communication through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.);

[0185] A bus for transmitting information between various components of the device (such as the processor 400, the memory 500, the input / output interface, and the communication interface);

[0186] Among them, the processor 400, the memory 500, the input / output interface, and the communication interface can achieve communication connections with each other inside the device through the bus.

[0187] The content in the above method embodiments is applicable to this robot embodiment. The functions specifically implemented in this robot embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those in the above method embodiments.

[0188] Thus, the embodiments of the present application have at least one of the following technical effects:

[0189] 1. Without relying on external perception devices such as lidar and force sensors, the dynamic model of the bipedal wheeled robot itself can accurately estimate the ground contact force without external perception, enabling the bipedal wheeled robot to more quickly and accurately perceive the terrain of the sole of the foot, allowing the bipedal wheeled robot to adaptively change the leg length based on the terrain and maintain the height stability of its head, thereby improving the control accuracy of the bipedal wheeled robot.

[0190] 2. Adopting an admittance control method to flexibly control the legs of the bipedal wheeled robot, enabling it to better adapt to complex terrains and maintain the height stability of its head. Through admittance control, the bipedal wheeled robot can exhibit dynamic characteristics similar to a spring-damper system when interacting with complex terrains. This characteristic allows the bipedal wheeled robot to freely adjust its motion state on surfaces of different hardnesses, making the actual leg length of the bipedal wheeled robot more tend to the desired leg length, better adapting to complex terrains, and improving the stability of its head under global control, thereby effectively improving the control accuracy of the bipedal wheeled robot.

[0191] 3. Introducing an adaptive parameter mechanism, using different admittance controller parameters for different terrains, enabling the bipedal wheeled robot to have better adaptability when facing different terrains and improving the stability of the bipedal wheeled robot during movement.

[0192] In summary, the embodiment of the present application is a set of head stability control schemes for a two-wheeled and legged robot based on admittance control. A simplified floating base model of the two-wheeled and legged robot is built, and the contact forces between the two wheels of the two-wheeled and legged robot and the ground are estimated according to the established dynamic model, so as to indirectly judge the ground state. Finally, the legs of the two-wheeled and legged robot are controlled through admittance control, enabling the head of the two-wheeled and legged robot to better adapt to the terrain. An adaptive parameter mechanism is introduced, which can better adapt to different terrains and has better disturbance rejection performance.

[0193] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order mentioned in the operating diagrams. For example, depending on the functions / operations involved, two consecutive blocks shown may actually be executed substantially simultaneously or the blocks can sometimes be executed in the reverse order. In addition, the embodiments presented and described in the flowcharts of the present application are provided by way of example for the purpose of providing a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logical flows presented herein. Alternative embodiments are contemplated, where the order of various operations is changed and where sub-operations described as part of a larger operation are executed independently.

[0194] Furthermore, although the present application has been described in the context of functional modules, it should be understood that, unless otherwise stated to the contrary, one or more of the functions and / or features may be integrated in a single physical device and / or software module, or one or more functions and / or features may be implemented in separate physical devices or software modules. It can also be understood that a detailed discussion of the actual implementation of each module is not necessary for understanding the present application. Rather, considering the attributes, functions, and internal relationships of the various functional modules in the devices disclosed herein, the actual implementation of the modules will be understood within the ordinary skills of an engineer. Therefore, those skilled in the art can implement the present application as set forth in the claims without undue experimentation. It can also be understood that the specific concepts disclosed are merely illustrative and are not intended to limit the scope of the present application, which is determined by the full scope of the appended claims and their equivalents.

[0195] If the above-mentioned functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several programs for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of this application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs.

[0196] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a sequenced list of executable programs for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by a program execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can retrieve and execute programs from a program execution system, apparatus, or device), or in conjunction with these program execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in conjunction with a program execution system, apparatus, or device.

[0197] More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection portion (electronic device) having one or more wirings, a portable computer disk cartridge (magnetic device), random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, a computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, then editing, interpreting, or otherwise processing it as appropriate, and then storing it in a computer memory.

[0198] It should be understood that various parts of the present application can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable program execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0199] In the above description of this specification, the description with reference to the terms "one embodiment / example", "another embodiment / example" or "certain embodiments / examples", etc. means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0200] Although the embodiments of the present application have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and purposes of the present application, and the scope of the present application is defined by the claims and their equivalents.

[0201] The above is a specific description of the preferred embodiments of the present application, but the present application is not limited to the described embodiments. Those skilled in the art can also make various equivalent deformations or substitutions without departing from the spirit of the present application, and these equivalent deformations or substitutions are all included within the scope defined by the claims of the present application.

Claims

1. A method for controlling the admittance of a two-wheeled foot robot, characterized in that: The following steps are involved: Performing admittance control based on the initial expected leg length data of each leg and the ground contact force to obtain the target expected leg length data of each leg; wherein the ground contact force is obtained by estimating the contact force according to the target motor torque of each leg in the previous control cycle, and the ground contact force in the first control cycle is zero; Using the target expected leg length data, the actual leg length data and other control data of each leg to perform torque processing, so as to obtain the target motor torque of each leg; Each of the legs is controlled according to a target motor torque of each of the legs.

2. The admittance control method of the two-wheeled leg robot according to claim 1, characterized in that: The method of performing admittance control based on the initial expected leg length data of each leg and the ground contact force to obtain the target expected leg length data of each leg includes: Based on the initial expected leg length data of each leg, first processing information of each leg is obtained; wherein the first processing information includes the change acceleration and change rate of the initial expected leg length data; According to the first processed information and the ground contact force of each leg, combined with a preset admittance control equation, the target expected leg length data of each leg is obtained.

3. The admittance control method of the two-wheeled foot robot according to claim 1, characterized in that: The target motor torque includes knee joint motor torque and other target motor torques; the target expected leg length data, actual leg length data and other control data of each leg are used for torque processing to obtain the target motor torque of each leg, including: Using the target expected leg length data and the actual leg length data of each leg to perform torque processing, so as to obtain the knee joint motor torque of each leg; The other control data of each leg is used to perform torque processing to obtain other target motor torques of each leg.

4. The admittance control method of the two-wheeled leg robot according to claim 3, characterized in that: The method of using the target expected leg length data and the actual leg length data of each leg to perform torque processing to obtain the knee joint motor torque of each leg includes: Based on the actual leg length data and the target expected leg length data of each leg, second processing information is obtained; wherein the second processing information includes the change acceleration and change rate of the target expected leg length data and the change rate of the actual leg length data; Obtaining a target output force of each leg according to the second processed information of each leg, the actual leg length data and the target expected leg length data; According to the target output force of each leg, the knee joint motor torque of each leg is obtained.

5. The admittance control method of the two-wheeled leg robot according to claim 1, characterized in that: The contact force estimation according to the target motor torque of each leg in the previous control cycle includes: According to the target motor torque of each leg in the previous control cycle, combined with the generalized coordinates of the two-wheeled foot robot, an equation of the joint space acceleration of each leg with respect to the ground contact force is obtained; Perform wheel rolling constraint processing on each leg to obtain the task space acceleration of the contact point between each leg and the ground; The ground contact force is obtained according to an equation of the joint space acceleration of each leg with respect to the ground contact force and the task space acceleration of the contact point between each leg and the ground.

6. The admittance control method of the two-wheeled leg robot according to claim 5, characterized in that: The equation of the joint space acceleration of each leg with respect to the ground contact force is obtained based on the target motor torque of each leg in the previous control cycle and the generalized coordinates of the two-wheeled foot robot, including: According to the target motor torque of each leg in the previous control cycle, combined with the generalized coordinates of the two-wheeled foot robot, a dynamic model of the two-wheeled foot robot is obtained; The mass matrix in the dynamic model is inverted to obtain an equation of the joint space acceleration of each leg with respect to the ground contact force.

7. The admittance control method of the two-wheeled foot robot according to claim 5, characterized in that: The performing wheel rolling constraint processing on each leg to obtain the task space acceleration of the contact point between each leg and the ground includes: Based on the generalized coordinates of the two-wheeled leg robot, the task space velocity of the contact point between each leg and the ground is obtained; The task space velocity of each contact point between the leg and the ground is derived to obtain the task space acceleration of each contact point between the leg and the ground.

8. The admittance control method of the two-wheeled leg robot according to claim 5, characterized in that: The ground contact force is obtained according to an equation of the joint space acceleration of each leg with respect to the ground contact force and the task space acceleration of the contact point between each leg and the ground, comprising: Determine the task space acceleration of the contact point between each leg and the ground to be zero, and obtain the ground rolling constraint state equation of each leg; Processing the ground rolling constraint state equation of each leg to obtain the joint space acceleration of each leg; The ground contact force is obtained according to the joint space acceleration of each leg and the equation of the joint space acceleration of each leg with respect to the ground contact force.

9. An admittance control device for a two-wheeled foot robot, characterized in that: include: An admittance controller, configured to perform admittance control based on initial desired leg length data of each leg and ground contact force, and obtain target desired leg length data of each leg; wherein the ground contact force is obtained by estimating the contact force according to the target motor torque of each leg in the previous control cycle, and the ground contact force in the first control cycle is zero; A torque processor, used for performing torque processing using the target expected leg length data of each leg, the initial expected leg length data and other control data to obtain a target motor torque of each leg; The leg controller is used to independently control each leg according to the target motor torque of each leg.

10. A two-wheeled foot robot, characterized in that: include: at least one processor; at least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements the admittance control method of the two-wheeled leg robot as described in any one of claims 1-8.

Images