Mechanical arm control method and equipment and computer readable storage medium

Abstract

According to the mechanical arm control method and device and the computer readable storage medium, when a mechanical arm executes a target contact task, data of the tail end contact force state and the joint state collected by a tail end sensor of the mechanical arm is obtained, and the actual situation of the mechanical arm during task execution is reflected. And the joint angular acceleration and the tail end contact force in the dynamic model are optimized in combination with a target contact task, so that the mechanical arm can dynamically adjust the joint angular acceleration and the tail end contact force according to different task requirements and environment changes. And more flexible self-adaptive contact control can be realized when the mechanical arm is in contact with the external environment through the optimized joint angular acceleration and the optimized tail end contact force. And finally, reasonable target joint torque is determined according to the optimized joint angular acceleration and the optimized tail end contact force, so that joint movement of the mechanical arm is controlled, and movement of the mechanical arm is more stable and reliable.

Description

Technical Field

[0001] The present application relates to the field of robot arm control technology, and in particular to a robot arm control method, device and computer-readable storage medium. Background Art

[0002] Multi-axis manipulators are widely used in the fields of mechanical processing, product manufacturing, and aerospace. They have a configuration composed of multiple joints. The combined motion between the joints enables the end tool to track a more complex trajectory in a three-dimensional space, and effectively perform delicate and complex tasks such as welding. Among them, the execution tasks of multi-axis manipulators can be divided into two types: static interaction and dynamic interaction according to the interaction with the outside world. Static interaction tasks are performed in a stable and standard working environment. In the process of performing such tasks, the manipulator does not need to actively adapt to changes in the environment, but mechanically executes a pre-specified trajectory. For dynamic interaction tasks, the working area of ​​the manipulator has great uncertainty, and it needs to be adjusted in time according to the environment of the task and changes in the environment to ensure the safety and reliability of the task execution. However, due to the limited perception ability of the current manipulator to the environment, the execution efficiency of dynamic interaction tasks is obviously inflexible, and in many cases it cannot even be compared with manual work, which also makes the application of manipulators in non-standard industrial scenarios subject to relatively large restrictions.

[0003] In order to enhance the adaptability of the robot arm to dynamic interaction tasks, the impedance control method is currently used to improve tactile perception and control capabilities. This method allows the robot arm to adjust its internal driving force according to changes in external forces to achieve more flexible contact interaction. However, the effect of impedance control depends largely on the accuracy of the control model. When the scene changes, such as when key factors such as the contact surface change significantly, traditional impedance control cannot adapt to such changes quickly and accurately due to its lack of adaptability, and cannot meet new task requirements well. Therefore, it is easy to cause overshoot instability in such contact problems. Summary of the invention

[0004] To overcome the problems existing in the related art, this specification provides a robot arm control method, device and computer-readable storage medium.

[0005] According to a first aspect of an embodiment of this specification, a method for controlling a robot arm is provided, the method comprising:

[0006] Acquiring data on the end contact force state and the joint state collected by the end sensor of the robotic arm when the robotic arm performs a target contact task;

[0007] According to the data of the end contact force state and the joint state, in combination with the target contact task, the joint angular acceleration and the end contact force in the dynamic model are optimized to obtain the optimized joint angular acceleration and the optimized end contact force corresponding to the end posture of the manipulator when it meets the expected trajectory when executing the target contact task;

[0008] Determine a target joint torque based on the optimized joint angular acceleration and the optimized end contact force;

[0009] According to the target joint torque, the joint movement of the robotic arm is controlled so that the movement trajectory of the end of the robotic arm is consistent with the expected trajectory when performing the target contact task.

[0010] According to a robot arm control method provided by the present application, the method further includes:

[0011] A quadratic optimization problem for the target contact task is constructed by combining the motion characteristics of the robotic arm when performing the target contact task and the dynamic model of the robotic arm; the quadratic optimization problem for the target contact task is used to obtain the corresponding optimized joint angular acceleration and optimized end contact force when the end posture of the robotic arm meets the expected trajectory when performing the target contact task.

[0012] According to a robot arm control method provided by the present application, the quadratic optimization problem is constructed by combining the motion characteristics of the target contact task and the dynamic model of the robot arm, including:

[0013] Modeling the robot arm based on the motion characteristics of the target contact task to obtain a contact model and corresponding contact constraints of the robot arm;

[0014] Acquire the dynamic model of the robotic arm; the dynamic model is used to represent the correlation between the joint torque of the robotic arm and the end contact force and the joint angular acceleration;

[0015] constructing a quadratic objective function according to the target contact task;

[0016] Taking the quadratic objective function as the optimization target, combining the contact model, the contact constraint and the dynamic model, a quadratic optimization problem of the target contact task is constructed.

[0017] According to a robot arm control method provided by the present application, the quadratic objective function includes at least one of the following:

[0018] A target tracking item for tracking and correcting the end state of the robotic arm;

[0019] A contact holding item for maintaining a base state of the end of the robot arm;

[0020] A contact force adjustment item for adjusting the end contact force of the robot arm.

[0021] According to a robot arm control method provided by the present application, at least one item in the quadratic objective function corresponds to a target weight;

[0022] The target weights of each item are adjusted to meet the control requirements of the end contact force state and the joint state when the robot arm performs the target contact task.

[0023] According to a robot arm control method provided by the present application, the joint angular acceleration and the end contact force in the dynamic model are optimized according to the data of the end contact force state and the joint state, combined with the target contact task, to obtain the corresponding optimized joint angular acceleration and the optimized end contact force when the end posture of the robot arm meets the expected trajectory when executing the target contact task, including:

[0024] Obtaining at least one of the joint torque, joint angular acceleration, and end contact force of the robotic arm;

[0025] The joint angular acceleration and / or the end contact force of the robot arm are optimized according to a constraint relationship associated with at least one of the joint torque, the joint angular acceleration, and the end contact force.

[0026] According to a robot arm control method provided by the present application, the optimization of the joint angular acceleration and / or the end contact force of the robot arm according to a constraint relationship associated with at least one of the joint torque, the joint angular acceleration, and the end contact force includes at least one of the following:

[0027] Optimizing the joint torque of the robotic arm so that the joint torque corresponding to the optimized target joint torque meets the constraint of the torque limit;

[0028] Optimizing the joint angular acceleration of the robotic arm so that the optimized joint angular acceleration meets the constraint of angular acceleration limit;

[0029] The end contact force of the robot arm is optimized so that the optimized end contact force satisfies the contact force constraint associated with the desired trajectory when the robot arm performs a target contact task.

[0030] According to a robot arm control method provided by the present application, the target joint torque is determined based on the optimized joint angular acceleration and the optimized end contact force, including:

[0031] Acquire the correlation between the joint torque of the robot arm and the end contact force and the joint angular acceleration;

[0032] The target joint torque is determined based on the association relationship, the optimized end contact force and the optimized joint angular acceleration.

[0033] The present application also provides a robotic arm control device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, any one of the robotic arm control methods described above is implemented.

[0034] The present application also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the robot arm control method as described in any one of the above is implemented.

[0035] The robot control method, device and computer-readable storage medium in the embodiment of this specification, compared with the problem that parameter overshoot and model uncertainty in the current impedance control technology easily lead to system instability, obtain the data of the end contact force state and joint state collected by the end sensor of the robot arm when the robot arm performs the target contact task, reflecting the actual situation of the robot arm when performing the task. Combined with the target contact task, the joint angular acceleration and end contact force in the dynamic model are optimized, so that the robot arm can dynamically adjust the joint angular acceleration and end contact force according to different task requirements and environmental changes, so that the end position of the robot arm more accurately meets the expected trajectory when performing the target contact task. And the optimized joint angular acceleration and end contact force can enable the robot arm to achieve more flexible adaptive contact control when contacting with the external environment, which helps to reduce the impact and damage risk to the contact object. Finally, the reasonable target joint torque is determined by the optimized joint angular acceleration and the optimized end contact force to control the joint movement of the robot arm, which can effectively avoid the system instability caused by parameter overshoot and model uncertainty, making the movement of the robot arm more stable and reliable.

[0036] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present specification. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the specification and, together with the description, serve to explain the principles of the specification.

[0038] Figure 1 is a flow chart of a method for controlling a robot arm according to an exemplary embodiment of the present specification;

[0039] Figure 2is another flowchart of a method for controlling a robotic arm according to an exemplary embodiment of the present specification;

[0040] Figure 3 is a schematic diagram of welding task contact according to an exemplary embodiment of the present specification;

[0041] Figure 4 is a schematic diagram of a robot arm control device according to an exemplary embodiment of the present specification;

[0042] Figure 5 It is a schematic block diagram of a robot arm control device shown in this specification according to an exemplary embodiment. DETAILED DESCRIPTION

[0043] Here, the technical solutions in the embodiments (or "implementations") of the present application will be described clearly and completely in conjunction with the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements.

[0044] If there are terms involving directional indications or positional relationships in the embodiments of the present application (such as up, down, left, right, front, back, inside, outside, top, bottom, center, vertical, horizontal, longitudinal, transverse, length, width, counterclockwise, clockwise, axial, radial, circumferential, etc.), such terms are only used to explain the relative positional relationship, movement, etc. between the components under a certain specific posture (as shown in the accompanying drawings); if the specific posture changes, the directional indication or positional relationship will also change accordingly. In addition, the terms "first", "second", etc. involved in the embodiments of the present application are only used for the purpose of convenience of description and cannot be understood as indicating or implying relative importance.

[0045] The present application provides a robot arm control method, device and computer readable storage medium. The present application is described in detail below in conjunction with the accompanying drawings. In the absence of conflict, the features of the following embodiments and implementations can be combined with each other.

[0046] In order to improve the adaptability of the robot arm to dynamic interactive tasks, the impedance control method is currently used to improve tactile perception and control capabilities. The traditional impedance control method is related to the joint torque, which assumes that the environmental contact force obtained by the end tool of the robot arm is expressed as a second-order spring damping system, so that the end position can be dynamically adjusted by measuring the end force, thereby achieving a compliant effect.

[0047] Usually, the impedance controller sets a fixed end stiffness or flexibility matrix, and uses the matrix as the control gain to perform a PD (Proportional-Derivative) style of control. Using this type of impedance controller will make the end behave like a spring with fixed stiffness, so that even if an external force is used to pull the end, the reaction force of the robot arm and the size of the posture remain linear.

[0048] Therefore, the effect of impedance control depends largely on the accuracy of the control model used. For fixed scenarios and tasks, impedance control can select the optimal stiffness and flexibility matrix by adjusting parameters. However, once the scenario changes, especially the contact surface changes, the performance of impedance control will deteriorate due to lack of adaptability, which can easily lead to overshoot instability in contact problems. For example, if the impedance control method suitable for metal operation is applied to rubber, the operation will produce relatively large deformation due to the excessive stiffness of the end, which will affect the execution efficiency.

[0049] Among them, overshoot refers to the excessive response of the robot arm to external force, which exceeds the expected range; instability means that the movement of the robot arm becomes unstable, and oscillation, jitter or uncontrollable movement may occur, which will seriously affect the normal operation and task execution of the robot arm, and even cause damage to the robot arm and the surrounding environment.

[0050] In order to solve the above technical problems, this specification provides a robot arm control method.

[0051] The present application provides an adaptive control method that can be used to improve the contact effect of a robotic arm. This method models the contact control problem of the robotic arm according to a dynamic model, uses the constraints provided by the model and the corresponding control objectives to transform the problem into a corresponding quadratic optimization problem, and uses the optimized output results as input to control the robotic arm. Compared with the traditional impedance control method, the method used in this application has lower requirements for the environmental model, and therefore has better adaptability, allowing the robotic arm to achieve smoother contact control. At the same time, since quadratic optimization has a mature solution algorithm, this application can achieve real-time control on ordinary embedded systems, thus having greater scene application value.

[0052] Next, the embodiments of this specification are described in detail.

[0053] Figure 1 is a flow chart of a method for controlling a robotic arm provided in an embodiment of this specification. Figure 1 Said method comprises the following steps:

[0054] In step 110, data of the end contact force state and the joint state collected by the end sensor of the robotic arm when the robotic arm performs the target contact task are obtained;

[0055] In step 120, according to the data of the end contact force state and the joint state, in combination with the target contact task, the joint angular acceleration and the end contact force in the dynamic model are optimized to obtain the optimized joint angular acceleration and the optimized end contact force corresponding to the end posture of the manipulator when it meets the expected trajectory when executing the target contact task;

[0056] At step 130, a target joint torque is determined based on the optimized joint angular acceleration and the optimized end contact force;

[0057] In step 140, the joint motion of the robotic arm is controlled according to the target joint torque so that the motion trajectory of the end of the robotic arm is consistent with the expected trajectory when performing the target contact task.

[0058] This embodiment aims to flexibly adjust the control strategy and resource allocation of the robot arm according to the task requirements when the robot arm performs the target contact task and the actual dynamic characteristics of the robot arm (for example, in different contact tasks, the robot arm can dynamically adjust the torque distribution of each joint according to factors such as the difficulty of the task and the requirements for flexibility to better complete the task). By adjusting specific internal parameters, the robot arm can show higher flexibility when in contact with the external environment, reducing the risk of impact and damage to the contact object. At the same time, by converting the adaptive contact control problem of the target contact task into a quadratic optimization problem with an optimization function as the target, and by adjusting the weights in the optimization function, the position control and contact force control of the end of the robot arm are flexibly balanced, so that the stiffness of the robot arm can be dynamically adjusted according to the task requirements, so that it can better adapt to different working conditions.

[0059] It is understandable that different task scenarios have different requirements for position accuracy and contact force. For example, in some tasks that require high-precision positioning, it may be necessary to pay more attention to position control, and the weight of position-related items in the objective function can be increased; while in some tasks that require flexible contact with the external environment, it may be necessary to pay more attention to contact force control, and the weight of contact force-related items can be increased.

[0060] Furthermore, the solution to the quadratic optimization problem is relatively insensitive to the weights of the optimization function, which means that when adjusting the weights for parameter adjustment, the solution will not fluctuate greatly due to slight changes in the weights, which makes the parameter adjustment process more stable and easier to obtain better control effects.

[0061] This embodiment solves the problem in the prior art impedance control technology that parameter overshoot and model uncertainty easily lead to system instability through the above-mentioned method, thereby achieving more flexible adaptive contact control.

[0062] As an example, the robot arm control method may be applied to a robot arm control system, which includes a robot arm and a control unit.

[0063] The robotic arm comprises a robotic arm body, a joint actuator and an end actuator.

[0064] Among them, various sensors are installed at various key positions on the robot body, such as sensors installed at the joints to measure the angles of the joints, force sensors installed on the end effectors of the robot arm to measure the force when grasping objects, etc. These sensors will feed back information such as the position, speed, and force of the joints to the control unit. The sensors are connected to the robot body through fixtures or interfaces to ensure that the sensors can work stably and accurately obtain data. For example, when the robot arm performs a target contact task, the end sensor collects data on the end contact force state and joint state, and provides the control system with real-time information about the contact between the robot arm and the external environment and the state of its own joints. These data are the basis for subsequent optimization and control.

[0065] The joint actuator converts the target joint torque into actual mechanical motion according to the target joint torque determined by the control system, adjusts the position and posture of each joint of the robotic arm to control the joint motion of the robotic arm, and makes the end motion trajectory of the robotic arm consistent with the expected trajectory when performing the target contact task.

[0066] The control unit includes hardware devices such as a central processing unit and a controller. The control unit adjusts the control instructions in real time according to these feedback information to achieve precise control of the robotic arm. For example, according to the data collected by the end sensor, combined with the target contact task, the joint angular acceleration and the end contact force in the dynamic model are optimized to obtain the optimized joint angular acceleration and the end contact force corresponding to the desired trajectory when the end posture of the robotic arm meets the target contact task. And based on the optimized joint angular acceleration and the end contact force, the target joint torque is determined. It can be understood that the control unit executes the control algorithm, and calculates the optimal control parameters (joint angular acceleration, end contact force and joint torque) by processing and optimizing the data to achieve precise control of the robotic arm.

[0067] As another example, the system includes a robotic arm, a control unit, and a host computer. The robotic arm and the control unit are consistent with the robotic arm and the control unit in the above example. The host computer is used to transmit information such as instructions and task parameters input by the user to the control unit of the robotic arm. At the same time, the controller will also transmit the state information, feedback data, etc. of the robotic arm back to the host computer, so that the user can perform operations such as task planning, monitoring and analysis on the host computer.

[0068] like Figure 1 and Figure 2 The above steps are described in detail as follows:

[0069] In step 110, data of the end contact force state and the joint state collected by the end sensor of the robotic arm when the robotic arm performs the target contact task is obtained.

[0070] The dynamic interaction task of the multi-axis robot is realized through its end tool. When the end of the robot comes into contact with the outside world, the robot will be subjected to the reaction force of the contact at the end of the tool, that is, the end contact force. The state of the end contact force will affect the movement of the robot. For example, if the end is subjected to a large contact force, this contact force will be transmitted to each joint through the structure of the robot, which may cause the joint angle of the robot, the size and direction of the joint torque to change to adapt to the effect of this force.

[0071] Therefore, when the robotic arm is performing a target contact task, the data of the end contact force state and joint state (including joint angle, joint velocity, joint angular acceleration and other information) collected by the sensors on the robotic arm are obtained. These data provide the current movement status of the robotic arm, so that the joint torque required to achieve the target contact task can be calculated based on the current movement status of the robotic arm.

[0072] In step 120, based on the data of the end contact force state and the joint state, combined with the target contact task, the joint angular acceleration and the end contact force in the dynamic model are optimized to obtain the corresponding optimized joint angular acceleration and the optimized end contact force when the end posture of the robot arm meets the expected trajectory when executing the target contact task.

[0073] The joint angular acceleration affects the motion state and speed change of the robot arm. Optimizing the joint angular acceleration can make the robot arm move more smoothly and accurately to the target posture, reducing jitter and error during movement. The end contact force is crucial for the robot arm to perform target contact tasks. Optimizing the end contact force can ensure that the contact between the robot arm and the external environment is in line with expectations, showing higher compliance and reducing the risk of impact and damage to the contact object.

[0074] Different target contact tasks and environments have different requirements and constraints on joint angular acceleration and end contact force. For example, grasping tasks require specific end contact force and appropriate joint motion trajectory, while assembly tasks may have higher requirements for position accuracy and contact force control. Therefore, combined with the optimization of target contact tasks, the robot arm can dynamically adjust the joint angular acceleration and end contact force according to different task requirements and environmental changes. Whether facing different contact surface characteristics, external interference or changes in task requirements, the robot arm can quickly adapt to ensure that the movement and contact of the robot arm meet the task requirements, ensuring stable and efficient operation.

[0075] The movement of the robot arm is constrained by the dynamic model, including factors such as joint inertia, friction, gravity, etc. Optimization combined with the dynamic model can make the optimization results more consistent with the actual movement law of the robot arm and improve the stability and reliability of the system.

[0076] In this embodiment, the target contact task and the dynamic model provide different information and constraints, and combining the two for optimization can achieve global optimization that comprehensively considers multiple factors. In this way, a better joint angle acceleration and end contact force can be obtained, thereby better controlling the robot arm to perform the target contact task.

[0077] The posture in this article is a general term for position and attitude. In the control of the robotic arm, accurately controlling the position and attitude of the end is crucial to completing various tasks (such as grasping objects, performing precise assembly operations, etc.). The posture can be divided into two-dimensional posture and three-dimensional posture. The two-dimensional posture includes horizontal position and vertical orientation, a total of three degrees of freedom information. The three-dimensional posture includes three-dimensional position and orientation, a total of six degrees of freedom information. In the present disclosure, the posture is a three-dimensional posture, specifically including the position information and orientation information of the end effector.

[0078] As an example, the robotic arm control method also includes: in step 140, a quadratic optimization problem of the target contact task is constructed in combination with the motion characteristics of the robotic arm when performing the target contact task and the dynamic model of the robotic arm; the quadratic optimization problem of the target contact task is used to obtain the corresponding optimized joint angular acceleration and optimized end contact force when the end posture of the robotic arm meets the expected trajectory when performing the target contact task.

[0079] The adaptive control problem of the robot arm is converted into a quadratic optimization problem, which is solved to obtain the robot arm motion state and contact force that meet the target contact task requirements, thereby achieving precise control of the robot arm.

[0080] In some embodiments, step 140 may specifically include: in step 141, modeling the robotic arm based on the motion characteristics of the target contact task to obtain the contact model and corresponding contact constraints of the robotic arm; in step 142, obtaining the dynamic model of the robotic arm; the dynamic model is used to represent the correlation between the joint torque of the robotic arm and the end contact force and joint angular acceleration; in step 143, constructing a quadratic objective function according to the target contact task; in step 144, taking the quadratic objective function as the optimization target, combining the contact model, the contact constraints and the dynamic model, to construct a quadratic optimization problem of the target contact task.

[0081] The process of creating a quadratic optimization problem for the target contact task involves three parts: modeling the robot contact task, creating a robot dynamics model, and constructing a quadratic optimization problem. A complete quadratic optimization problem is formed to optimize the joint angular acceleration and end contact force.

[0082] The modeling process of the robot arm contact task is as follows:

[0083] The dynamic interaction task of the multi-axis robot is realized through its end effector, which is directly related to the coordinate position and posture of the end effector. We use x′=[x, y, z] to represent the position of the end effector in the world coordinate system, use the three-dimensional rotation matrix R′∈SO(3) to represent the posture of the tool coordinate system in the world coordinate system, and use q to represent the angles of each joint of the robot. Among them, SO(3) is a Lie group set of third-order rotation matrices, which represents the posture orientation of the rigid body, that is, each element can represent a certain posture of the rigid body.

[0084] Then, according to the contact model of the robot arm, the end position and posture can be represented by a function, and exemplarily represented as follows:

[0085] x′=x′(q)=[x(q),y(q),z(q)],R′=R′(q)

[0086] On the other hand, the contact constraints of the dynamic interaction process of the target contact task can be expressed as:

[0087] F(x′,R′)=F(x′(q),R(q))=F * (q) = f(t)

[0088] G(x′,R′)=G(x(q),R(q))=G * (q) = 0

[0089] Among them, the function F(x′, R′) represents the relevant quantity that the task needs to track, and G(x′, R′) represents the contact state that the task needs to maintain. Both are vector functions. * (q) is the expression of the target contact task in the joint angle space, which can be used to construct constraints related to target tracking of the target contact task. * The specific expression of (q) can be obtained by substituting the analytical expression of the end state of the robot arm (which can be a mathematical expression describing the relationship between the end position, posture, etc. of the robot arm and the joint angle) into the contact function F.

[0090] G*(q) is the potential function of the contact task in the joint angle space. In order to ensure that the contact remains unchanged, the value of the potential function needs to be kept constant, which can be used to form the contact-maintaining related constraints of the target contact task. The specific expression of G*(q) can be obtained by substituting the analytical expression of the end state of the robot (which can be a mathematical expression describing the relationship between the end position, posture, etc. of the robot and the joint angle) into the contact function G.

[0091] f(t) represents the desired position and orientation that can change over time. When the end effector of the robot arm performs the target contact task, there is an ideal position and orientation change pattern, which remains constant or changes dynamically over time. The actual motion trajectory of the end of the robot arm (including position and orientation) needs to be adjusted according to the contact situation, so that during the contact process, the end of the robot arm can meet the task requirements and adapt to the constraints brought by the contact. Therefore, for contact problems, due to the existence of contact, the actual trajectory of the end of the robot arm is consistent with the contact, that is, in the case of contact, the movement of the robot arm is an adjustment and adaptation to the ideal trajectory after considering the contact factor.

[0092] As an example, Figure 3 As shown, taking welding as an example, in order to ensure the mechanical strength of the welding surface, the welding gun at the end of the robot arm needs to maintain a specific posture and move along the contact surface until the weld is completely covered. The corresponding contact model is exemplarily represented as:

[0093]

[0094] G(x′,R′)=z=0

[0095] Among them, R d is the constant value of the third-order rotation matrix, which indicates the ideal posture that the end effector of the robot arm needs to achieve. It limits the rotation state that the robot arm is expected to achieve when performing the target contact task. This parameter does not change with time. x (t) represents the reference position trajectory of the end effector along the x-axis, where the parameter (t) indicates that the position changes with time.y (t) represents the reference position trajectory of the end effector along the y-axis direction, where the parameter (t) indicates that the position changes with time.

[0096] The above contact model indicates that the height and posture of the end of the robot arm are kept unchanged, and its horizontal position is moved so that it can track a specific trajectory.

[0097] Based on the above two functions related to the target contact task, we can model the robot arm and set contact constraints related to the contact force.

[0098] The process of creating a robotic arm dynamics model is as follows:

[0099] Assuming that a multi-axis robot needs to maintain n contacts, that is, the function G(x′, R′) is an n-dimensional vector, the dynamic model of the multi-axis robot is as follows:

[0100]

[0101] Among them, the variable q represents the vector of the robot arm joint angle, The vector representing the angular velocity of the robot joints, represents the vector of the angular acceleration of the robot joint, τ represents the vector of the robot joint torque, and f i Indicates the magnitude of the corresponding contact force or moment.

[0102] Other terms in the kinetic model have physical meanings: is the contact-related Jacobian matrix, which represents the influence of the contact constraint force of the robot on the dynamics of the robot. It is understood that since the robot arm is in contact with the external environment or the task object when performing the target contact task, the end effector of the robot arm will be subjected to the reaction force of the contact. The dynamic influence of this force on the robot arm needs to be expressed through the contact function G * (q). Taking ground contact as an example, the ground will continuously provide a support force on the end effector, which will affect the overall dynamics of the robot arm. Specifically, the reaction force is in the form of G * (q) Partial derivatives with respect to each joint angle (a row in the Jacobian matrix).

[0103] M(q) is the mass matrix, which represents the mass distribution of the robot arm. is the Coriolis force matrix, which represents the non-inertial coupling relationship of the robot arm. is the offset term, which represents the effect of gravity and internal friction of the manipulator.

[0104] According to the above dynamic model, it can be known that the external end contact force f iand joint angular acceleration When it is known, we can get the corresponding joint torque as:

[0105]

[0106] Conversely, it can also be shown that the robot arm can control the external contact force by adjusting the joint torque within a certain range. For the target contact task, the corresponding constraint force should satisfy the condition f i =N i (t)(i=1,…,n), each type of contact will generate corresponding contact force or contact torque, N i (t) refers to the expected trajectory of the force or torque corresponding to the i-th contact, so that the target contact task can be better executed. For example, in order to ensure that the screw is fixed well enough, it is necessary to maintain a certain torque and normal pressure during the tightening process. This torque and normal pressure are the expected trajectory of the force or torque, which serves as the goal of optimizing the end contact force and joint angular acceleration.

[0107] Continue to refer to Figure 3 , the corresponding constraint condition is f z =f d ,

[0108] Among them, f d Indicates the target normal pressure, that is, the contact support force is kept at a specific value. In the manipulator task, keeping the specific support force within a certain range can effectively improve the success rate of task execution; M d It represents the target contact torque, that is, the contact torque is maintained at a specific value. For a specific welding task, among the contact torques in the three directions, the torque in the Z-axis direction needs to be small enough to effectively avoid welding failure caused by surface stress concentration.

[0109] That is, the above constraint conditions indicate that the normal contact support force and contact torque are kept near specific values ​​to ensure the welding effect.

[0110] The process of constructing a quadratic optimization problem is as follows:

[0111] Combining the above two processes, the contact task goal of the multi-axis manipulator is to control the joint torque input of the manipulator so that the position and posture of the end can satisfy F(x′, R′) = f(t) as much as possible, while satisfying the dynamic equation and the corresponding contact constraint f i =N i (t).

[0112] To this end, a quadratic objective function is constructed according to the target contact task. The quadratic objective function is used to make the end position and posture of the robot arm as close to the desired state as possible by adjusting the joint torque input under the premise of satisfying the robot arm dynamics equations and contact constraints.

[0113] As an example, the quadratic objective function includes at least one of a target tracking item for tracking and correcting the end state of the robot arm; a contact maintenance item for maintaining the basic state of the end of the robot arm; and a contact force adjustment item for adjusting the contact force of the end of the robot arm.

[0114] In the first part, the desired state of the end of the robot arm is determined according to the target contact task, including the desired position, desired posture, etc. The error between the actual state and the desired state of the end is defined, and the target tracking item is constructed.

[0115] Exemplarily, it is represented as follows:

[0116]

[0117] The target tracking item of the target contact task is used to track and correct the terminal state. Represents the Jacobian matrix of the task tracking function with respect to the joint angle of the robot arm, that is, the first-order partial derivative. Represents the second-order partial derivative tensor of the task tracking function with respect to the joint angles of the robot arm. Function The purpose of the tracking feedback function is to achieve tracking by dynamically adjusting the state of the terminal.

[0118] Continue to refer to Figure 3 , PD control feedback can be used to track the horizontal position of the end, as shown below:

[0119]

[0120] Where x and y represent the position of the end effector in the x and y axis directions, and represents the velocity of the end effector in the x and y axis directions. x and f y represents the target position of the end in the x and y axis directions, f′ x and f′ y Represents the target speed in the x and y axis directions. k px and k dx represents the proportional control and differential control gain of the x-axis position; k py and k dy Represents the proportional and derivative control gains for the y-axis position.

[0121] Typically, an appropriate weight λ1 is set for the target tracking item to ensure that the deviation between the terminal state and the desired state is minimized during the optimization process.

[0122] The second part is to analyze the contact constraints and determine the basic state that the end of the robot arm needs to maintain during the contact process (such as maintaining a certain contact force, maintaining a stable contact position, etc.). According to the requirements of the contact constraints, define the indicators of the contact maintenance items. For example, if a specific contact force is required, the contact force deviation is defined as the difference between the actual contact force and the expected contact force. Then construct the contact maintenance items based on the above content.

[0123] Exemplarily, it is represented as follows:

[0124]

[0125] The contact maintenance item of the target contact task is used to maintain the current contact state to avoid contact switching.

[0126] Usually, an appropriate weight λ2 is set for the contact maintenance term to ensure that the contact constraints are satisfied as much as possible during the optimization process.

[0127] Compared with general impedance control, the constraint maintenance effect of this part is relatively gentle, that is, when the actual contact surface and the model function differ greatly, the effect of this part is relatively low, and it will not cause the robot to produce dangerous behaviors due to model errors, such as squeezing the contact surface with a relatively large force. According to the relevant knowledge of contact dynamics, in order to keep the contact unchanged, the robot needs to adjust the contact function G * (q) is differentiated twice to obtain the following second-order constraint equation:

[0128]

[0129] in, is a second-order tensor that satisfies is the property of the matrix. In order to maintain contact, the above conditions must be met, but due to the uncertainty of the external environment, the uncertainty of the contact function itself is also relatively large. Strictly maintaining the constraints will amplify the uncertainty of the model and affect the safety of the entire system. For example, if the end of the robot moves on an uneven surface, if the constraints are strictly maintained, the robot will get stuck directly in a relatively rough place, resulting in the insecurity of the entire system. Therefore, in the method of this application, the constraint maintenance term is added to the quadratic objective function, that is, This item may not be zero under certain circumstances. This soft constraint requirement also ensures the flexibility during the contact process and improves adaptability.

[0130] The third part is to determine the contact force range and target that needs to be adjusted. Define the indicators for contact force adjustment. The effect of contact force adjustment can be measured by indicators such as the change in actual contact force and the smoothness of contact force. Then, the contact force adjustment items are constructed based on these contents.

[0131] Exemplarily, it is represented as follows:

[0132]

[0133] The contact force adjustment item of the target contact task is used to adjust the contact force according to the feedback of the force sensor. i (t),f im ) is the contact feedback function, which aims to use the data of the force sensor and the target contact value to correct the end contact force in response to external changes and fluctuations. Among them, the target contact value refers to the expected trajectory of the contact force, f im Represents the measured value of the i-th contact force obtained by the force sensor.

[0134] Typically, an appropriate weight λ3 is set for the contact force adjustment term to encourage smooth contact force adjustment during the optimization process.

[0135] Continue to refer to Figure 3 , the correction of the end contact force can be expressed as:

[0136] H2(N i (t),f im )=N i (t)+ρ(f im -N i (t))

[0137] That is, the weighted average of the target value and the measured value is used for adjustment. This method is an IIR filtering method, which is used to achieve a relatively effective low-pass filter, where ρ represents the bandwidth of the filter. The smaller the value, the smaller the bandwidth of the filter, and the smoother the output reference value. On the contrary, a higher ρ value will cause the output reference value to be more susceptible to fluctuations, but the tracking performance is better.

[0138] Due to external uncertainties, there will be a gap between the actual contact force and the reference value. At this time, the robot arm needs to dynamically adjust the contact to perform reference tracking. For example, when the actual measured normal support force (such as 3N) is inconsistent with the reference support force (such as 5N) during ground contact, the robot arm needs to increase the pressure on the ground force to approach the reference support; on the other hand, in order to improve compliance, the process of increasing the pressure of the robot arm needs to be executed relatively smoothly, so using an intermediate value (such as 4N) for transition is a relatively good choice.

[0139] Thus, the quadratic objective function of the quadratic optimization problem constructed is exemplarily expressed as follows:

[0140]

[0141] Among them, H1 and H2 are auxiliary functions used for tracking. λ1 represents the weight of position tracking, λ2 represents the weight of contact maintenance, and λ3 represents the weight of contact force tracking. Their values ​​need to be adjusted according to the task. im Represents the magnitude of the i-th measured contact force or torque, which is obtained by the torque sensor at the end.

[0142] It can be understood that the target weights of each item are adjusted to meet the control requirements of the end contact force state and the joint state when the robot arm performs the target contact task. In other words, the relative importance of each item is adjusted by setting the weight coefficient, thereby limiting the role of the quadratic objective function in different aspects. For example, the weight of the target tracking item can be increased to emphasize the precise control of the end posture.

[0143] Finally, taking the quadratic objective function as the optimization target, combining the contact model, the contact constraint and the dynamic model, a quadratic optimization problem of the target contact task is constructed. The quadratic optimization problem is to find the variable value that makes the quadratic objective function obtain the optimal value (maximum value or minimum value) under certain constraints.

[0144] The constraints may include kinematic constraints of the robot arm (such as joint angle range, end effector position constraints, etc.) and dynamic constraints (such as joint torque limits, etc.).

[0145] As an example, step 140 also includes: obtaining at least one of the joint torque, joint angular acceleration, and end contact force of the robotic arm; and optimizing the joint angular acceleration and / or end contact force of the robotic arm based on a constraint relationship associated with at least one of the joint torque, the joint angular acceleration, and the end contact force.

[0146] The optimization of the joint angular acceleration and / or the end contact force of the robotic arm according to the constraint relationship associated with at least one of the joint torque, the joint angular acceleration, and the end contact force includes at least one of the following: optimizing the joint torque of the robotic arm so that the joint torque corresponding to the optimized target joint torque meets the constraint of torque limit; optimizing the joint angular acceleration of the robotic arm so that the optimized joint angular acceleration meets the constraint of angular acceleration limit; optimizing the end contact force of the robotic arm so that the optimized end contact force meets the contact force constraint condition related to the expected trajectory when the robotic arm performs the target contact task.

[0147] Specifically, first of all, it can be understood that the joint torque of the robot arm needs to be kept within a certain range. In view of this, the present application proposes a torque limit constraint for the joint torque of the robot arm. Exemplarily, the target joint torque of the robot arm satisfies the following relationship:

[0148]

[0149] The comparison symbol used in the torque amplitude constraint indicates that the absolute value of each component in the vector is within a specific range. The joint torque is limited according to the joint motor of the robot arm and the actual load condition to prevent the motor of the robot arm from being overloaded.

[0150] Furthermore, it is understandable that the joint angular acceleration of the robotic arm cannot be too large to ensure smooth movement of the robotic arm and reduce shock and vibration.

[0151] In view of this, the present application proposes a constraint on the angular acceleration limit of the joint angular acceleration of the robotic arm. Exemplarily, the target joint torque of the robotic arm satisfies the following relationship:

[0152]

[0153] The angular acceleration limit constraint is to avoid sudden jitter of the robot arm caused by excessive angular acceleration and to improve the smoothness of the overall control.

[0154] In addition, it can be understood that the end contact force and torque of the robot arm are within a safe range, avoiding damage to the contact object or affecting the stability of the robot arm, and satisfying the expected trajectory when the robot arm performs the target contact task.

[0155] In view of this, the present application proposes a contact force constraint for the end contact force of the robot arm. Exemplarily, the end contact force of the robot arm satisfies the following relationship:

[0156] f imin ≤f i ≤f imax

[0157] The contact force constraint ensures that the contact force and torque between the robot and the contact object are within a safe range to avoid damage to the contact object or affecting the stability of the robot. For example, the contact force constraint can prevent the robot from damaging the object by applying too much force when grasping the object; the contact torque constraint can ensure that the robot does not rotate or tilt unstably when in contact with the external environment.

[0158] Through the combined effect of these constraints, the robot arm can move safely, accurately, and efficiently when performing tasks, while meeting the task requirements and physical limitations. By setting these constraints reasonably, an effective quadratic optimization problem can be constructed to solve the optimal robot arm joint parameters and contact force and other optimization variables.

[0159] Exemplarily, the quadratic optimization problem is constructed as follows:

[0160]

[0161] The corresponding constraints are:

[0162]

[0163] f imin ≤f i ≤f imax

[0164] By adjusting the weights of each item of the quadratic objective function and optimizing this quadratic objective function, the joint state of the robot arm (including joint angular acceleration, etc.) and the end contact force that meet the requirements of the target contact task can be obtained, thereby achieving optimal control of the robot arm.

[0165] The variables obtained by solving the above quadratic optimization problem are That is, the optimized joint angular acceleration and the optimized end contact force.

[0166] In step 130 , the target joint torque is determined based on the optimized joint angular acceleration and the optimized end contact force.

[0167] The joint torque is a direct means of controlling the motion of the robot arm. By calculating the target joint torque and applying it to the joint actuator of the robot arm, the motion state and position of the robot arm are controlled. It can be understood that by calculating the target joint torque in real time and adjusting it according to the end contact force state, joint state and target contact task, the robot arm can quickly adapt to changes in tasks and environment. For example, when the contact surface changes, the robot arm can adjust the joint torque to maintain the appropriate contact force and achieve adaptive control. This can improve the adaptability of the robot arm and overcome the problem that traditional impedance control lacks adaptability when facing changes.

[0168] In some embodiments, step 130 may specifically include: in step 131, obtaining the correlation relationship between the joint torque of the robotic arm and the end contact force and the joint angular acceleration; in step 132, determining the target joint torque based on the correlation relationship, the optimized end contact force and the optimized joint angular acceleration.

[0169] From the above-mentioned dynamic model of the robot arm, it can be seen that there is a correlation between the joint torque of the robot arm and the end contact force and joint angular acceleration. The robot arm can control the external contact force by adjusting the joint torque within a certain range. Therefore, the calculation formula of the target joint torque is given as follows:

[0170]

[0171] in, This is the result of quadratic optimization. Other values ​​can be calculated through the dynamic model and joint states.

[0172] Since the above quadratic optimization always has a solution, it can be quickly solved using mature interior point or simplex numerical methods. Taking the seven-axis robot arm as an example, the time consumed for solving the above quadratic form on a general embedded system will not exceed one millisecond, so it can be solved quickly to achieve real-time control.

[0173] In this embodiment, by optimizing the key parameters in the dynamic model and determining the reasonable target joint torque, the system instability caused by parameter overshoot and model uncertainty can be effectively avoided. The movement of the robot arm is more stable and reliable, the probability of failure is reduced, and the overall stability of the system is improved.

[0174] In step 140, the joint motion of the robotic arm is controlled according to the target joint torque so that the motion trajectory of the end of the robotic arm is consistent with the expected trajectory when performing the target contact task.

[0175] The joint torque ultimately controls the joints of the robot arm. By applying torque to the joints, the motion state of the joints can be changed, which is then transmitted through the connecting rod structure of the robot arm to affect the position and posture of the end of the robot arm, so as to achieve control of the end of the robot arm to perform specific tasks.

[0176] Based on the above embodiments, a specific example of a robot arm control method is provided.

[0177] Taking the contact experiment process of the seven-axis robot arm as an example, the overall experimental purpose is for the robot arm to use the brush at the end to perform a specific trajectory on the desktop, thereby achieving the corresponding cleaning task. For example, if the specific trajectory is a periodic circular trajectory with (0.5, 0) as the center, 0.5m as the radius, 12 seconds as the period, and the origin as the initial position, the height of the contact desktop is 0.2m, and the brush at the end of the robot arm tracks the arc trajectory on the desktop while keeping the brush in contact with the desktop. Then the task tracking and contact maintenance functions in the specific contact and task can be specifically expressed as the following formula:

[0178]

[0179] G(x,R)=z-0.2=0

[0180] That is, the end tool posture is kept downward, and the end presses the brush to move in an arc. Since there is only one contact force for this target contact task, that is, the normal support force of the desktop. Assuming that the normal support force is 10N (Newton), the corresponding support force reference trajectory is:

[0181] N1(t)=10

[0182] That is, the normal support force is maintained at 10N during the contact process.

[0183] For the target tracking item of the end position, the relevant data of the target contact task is brought in. This application uses the following position tracking function (it should be noted that for the quadratic objective function given above, the expressions related to posture (joint angle q, joint angular velocity and joint angular acceleration ) Since it involves posture-related knowledge, its specific implementation method is to bring the numerical value of the posture-related expression into the posture-related expression in the quadratic objective function for calculation. Since the related expressions are long and complicated, no specific examples are given here, and they are expressed as posture-related items):

[0184]

[0185] For the contact force adjustment item at the end, this application uses the following contact force tracking function:

[0186] H2(N i (t),f im )=0.5N i (t)+0.5f im =5+0.5f im (ρ=0.5)

[0187] The corresponding QP solution problem is expressed as follows (where λ1=500; λ2=500; λ3=200 is set):

[0188]

[0189] According to the relevant indicators of the robotic arm, the corresponding constraints are:

[0190]

[0191] Finally, the quadratic optimization problem of the seven-axis robotic arm when performing the target contact task is solved to obtain the optimized joint angular acceleration and the optimized end contact force. Based on the optimized joint angular acceleration and the optimized end contact force, the target joint torque is determined. According to the target joint torque, the joint movement of the robotic arm is controlled so that the robotic arm uses the brush at the end to execute a specific trajectory on the desktop.

[0192] The present application provides a method, device and computer-readable storage medium for controlling a manipulator. Compared with the problem that parameter overshoot and model uncertainty in the current impedance control technology easily lead to system instability, the data of the end contact force state and joint state collected by the end sensor of the manipulator when the manipulator performs the target contact task are obtained to reflect the actual situation of the manipulator when performing the task. Combined with the target contact task, the joint angular acceleration and the end contact force in the dynamic model are optimized, so that the manipulator can dynamically adjust the joint angular acceleration and the end contact force according to different task requirements and environmental changes, so that the end position of the manipulator more accurately meets the expected trajectory when performing the target contact task. And the optimized joint angular acceleration and the end contact force can enable the manipulator to achieve more flexible adaptive contact control when contacting with the external environment, which helps to reduce the impact and damage risk to the contact object. Finally, a reasonable target joint torque is determined by the optimized joint angular acceleration and the optimized end contact force to control the joint movement of the manipulator, which can effectively avoid the system instability caused by parameter overshoot and model uncertainty, making the movement of the manipulator more stable and reliable.

[0193] Based on the same application concept as the above method, the present application embodiment also proposes a robot arm control device, such as Figure 4 shown.

[0194] The device comprises:

[0195] The data acquisition module 402 is used to acquire the data of the end contact force state and the joint state collected by the end sensor of the robotic arm when the robotic arm performs the target contact task;

[0196] A parameter optimization module 404 is used to optimize the joint angular acceleration and the end contact force in the dynamic model according to the data of the end contact force state and the joint state in combination with the target contact task, so as to obtain the optimized joint angular acceleration and the optimized end contact force corresponding to the end posture of the manipulator when the end posture meets the expected trajectory when executing the target contact task;

[0197] A torque determination module 406, configured to determine a target joint torque based on the optimized joint angular acceleration and the optimized end contact force;

[0198] The contact control module 408 is used to control the joint movement of the robotic arm according to the target joint torque so that the movement trajectory of the end of the robotic arm is consistent with the expected trajectory when performing the target contact task.

[0199] The implementation process of the functions and effects of each module / submodule / unit in the above-mentioned device is specifically described in the implementation process of the corresponding steps in the above-mentioned method, and the same technical effect can be achieved, so it will not be repeated here.

[0200] Figure 5 The following is a schematic diagram of the physical structure of a robotic arm control device. Figure 5 As shown, the robot arm control device may include: a processor 810, a communication interface 820, a memory 830 and a communication bus 840, wherein the processor 810, the communication interface 820 and the memory 830 communicate with each other through the communication bus 840. The processor 810 may call the logic instructions in the memory 830 to execute the robot arm control method.

[0201] In addition, the logic instructions in the above-mentioned memory 830 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art, and the computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0202] On the other hand, the present application also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the robot arm control methods provided by the above methods.

[0203] On the other hand, the present application also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the robot arm control method provided by the above-mentioned methods.

[0204] It should be noted that the technical solutions or technical features described in the above embodiments can be combined or supplemented with each other without causing conflicts. The scope of protection of this application is not limited to the precise structures described in the above embodiments and shown in the drawings; all modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of this application should be included in the scope of protection of this application.

Claims

1. A robot arm control method, characterized in that: The method comprises: Acquiring data on the end contact force state and the joint state collected by the end sensor of the robotic arm when the robotic arm performs a target contact task; According to the data of the end contact force state and the joint state, in combination with the target contact task, the joint angular acceleration and the end contact force in the dynamic model are optimized to obtain the optimized joint angular acceleration and the optimized end contact force corresponding to the end posture of the manipulator when it meets the expected trajectory when executing the target contact task; Determine a target joint torque based on the optimized joint angular acceleration and the optimized end contact force; According to the target joint torque, the joint movement of the robotic arm is controlled so that the movement trajectory of the end of the robotic arm is consistent with the expected trajectory when performing the target contact task.

2. The robot arm control method according to claim 1, characterized in that: The method further comprises: A quadratic optimization problem for the target contact task is constructed by combining the motion characteristics of the robotic arm when performing the target contact task and the dynamic model of the robotic arm; the quadratic optimization problem for the target contact task is used to obtain the corresponding optimized joint angular acceleration and optimized end contact force when the end posture of the robotic arm meets the expected trajectory when performing the target contact task.

3. The robot arm control method according to claim 2, characterized in that: The method of constructing a quadratic optimization problem by combining the motion characteristics of the target contact task and the dynamics model of the robotic arm includes: Modeling the robot arm based on the motion characteristics of the target contact task to obtain a contact model and corresponding contact constraints of the robot arm; Acquire the dynamic model of the robotic arm; the dynamic model is used to represent the correlation between the joint torque of the robotic arm and the end contact force and the joint angular acceleration; constructing a quadratic objective function according to the target contact task; Taking the quadratic objective function as the optimization target, combining the contact model, the contact constraint and the dynamic model, a quadratic optimization problem of the target contact task is constructed.

4. The robot arm control method according to claim 3, characterized in that: The quadratic objective function includes at least one of the following: A target tracking item for tracking and correcting the end state of the robotic arm; A contact holding item for maintaining a base state of the end of the robot arm; A contact force adjustment item for adjusting the end contact force of the robot arm.

5. The robot arm control method according to claim 4, characterized in that: At least one item in the quadratic objective function corresponds to a target weight; The target weights of each item are adjusted to meet the control requirements of the end contact force state and the joint state when the robot arm performs the target contact task.

6. The robot arm control method according to claim 1, characterized in that: The method optimizes the joint angular acceleration and the end contact force in the dynamic model according to the data of the end contact force state and the joint state in combination with the target contact task, and obtains the corresponding optimized joint angular acceleration and the optimized end contact force when the end posture of the manipulator meets the expected trajectory when executing the target contact task, including: Obtaining at least one of the joint torque, joint angular acceleration, and end contact force of the robotic arm; The joint angular acceleration and / or the end contact force of the robot arm are optimized according to a constraint relationship associated with at least one of the joint torque, the joint angular acceleration, and the end contact force.

7. The robot arm control method according to claim 6, characterized in that: The step of optimizing the joint angular acceleration and / or the end contact force of the robot arm according to a constraint relationship associated with at least one of the joint torque, the joint angular acceleration, and the end contact force comprises at least one of the following: Optimizing the joint torque of the robotic arm so that the joint torque corresponding to the optimized target joint torque meets the constraint of the torque limit; Optimizing the joint angular acceleration of the robotic arm so that the optimized joint angular acceleration meets the constraint of angular acceleration limit; The end contact force of the robot arm is optimized so that the optimized end contact force satisfies the contact force constraint associated with the desired trajectory when the robot arm performs a target contact task.

8. The robot arm control method according to claim 1, characterized in that: The step of determining the target joint torque based on the optimized joint angular acceleration and the optimized end contact force comprises: Acquire the correlation between the joint torque of the robot arm and the end contact force and the joint angular acceleration; The target joint torque is determined based on the association relationship, the optimized end contact force and the optimized joint angular acceleration.

9. A robot arm control device, characterized in that: It includes a memory, a processor, and a robot arm control program stored in the memory and executable on the processor, and when the processor executes the robot arm control program, the steps of the robot arm control method as described in any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a robot control program, which, when executed, implements the steps of the robot control method according to any one of claims 1 to 8.

Images