An Adaptive Force Adjustment Method for a Robotic Arm Based on Gravity Compensation and Impedance Control
By constructing a gravity compensation model and an impedance-controlled adaptive force adjustment method for the robotic arm, the problem of the independence between trajectory planning and force control in the collaborative handling of dual-arm robotic arms was solved. This improved the stability and operational performance of the robotic arm in complex environments and ensured the force balance and operational accuracy of the dual-arm collaborative control.
Patent Information
- Application Number
- CN202611011678.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-08
- Publication Date
- 2026-11-17
AI Technical Summary
In existing technologies for collaborative handling tasks using dual-arm robotic arms, trajectory planning and force control are independent of each other, lacking a unified collaborative optimization mechanism. This makes it difficult to cope with complex environments, resulting in insufficient force feedback or system instability. Furthermore, the lack of an effective force coordination mechanism during dual-arm collaboration makes it difficult to ensure force balance and operational accuracy during handling.
An adaptive force adjustment method for a robotic arm based on gravity compensation and impedance control is adopted. By constructing a gravity compensation model, the zero-point force of the sensor and the gravity of the end tool are calculated to generate the desired force. The state of the driven arm is adjusted through an impedance control model to achieve coordinated movement of the active arm and the driven arm.
It improves the stability and compliance of the robotic arm force control, enhances the consistency and coordination of dual-arm collaborative control, reduces the complexity of system control, and ensures force balance and operational accuracy during the handling process.
Smart Images

Figure CN123103618A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of robotic arm control technology, specifically to an adaptive force adjustment method for robotic arms based on gravity compensation and impedance control. Background Technology
[0002] Currently, research on collaborative handling tasks using dual-arm robotic arms mainly focuses on trajectory planning, force control, and human-machine collaborative control. In trajectory planning, traditional methods often employ offline planning strategies based on interpolation or optimization. While these can generate smooth trajectories, they exhibit poor adaptability to dynamic environments or task changes. Regarding force control, existing methods mostly employ fixed-gain force control or impedance control strategies, which can achieve compliant operation to some extent. However, due to fixed parameters, they struggle to adapt to different loads, objects with varying stiffness, and complex contact states, easily leading to insufficient force feedback or system instability. In dual-arm collaborative control, existing research often achieves coordination between the two arms through master-slave control or rigid constraints, but this often neglects force distribution and dynamic balance between the arms, resulting in uneven force distribution, object tilting, or even slippage during handling.
[0003] The existing technologies generally have the following shortcomings: (1) Trajectory planning and force control are independent of each other and lack a unified collaborative optimization mechanism, making it difficult to cope with complex interactive tasks. (2) There is a lack of effective force coordination mechanism in the process of dual-arm collaboration, making it difficult to ensure force balance and operational accuracy during the handling process.
[0004] In summary, there is an urgent need for a dual-arm collaborative control method that integrates end-effector force compensation and trajectory planning, and has adaptive impedance adjustment capabilities, in order to improve the stability, robustness, and operational performance of dual-arm handling tasks in complex environments. Summary of the Invention
[0005] To overcome the problems existing in related technologies, the purpose of this invention is to provide an adaptive force adjustment method for a robotic arm based on gravity compensation and impedance control. This method integrates end-effector force compensation and trajectory planning, and has an adaptive impedance adjustment function to improve the stability, robustness and operational performance of dual-arm handling tasks in complex environments.
[0006] An adaptive force adjustment method for a robotic arm based on gravity compensation and impedance control includes: When the robotic arm does not interact with the external environment, a gravity compensation model is constructed; Collect n end-effector force data of the robotic arm in different poses, substitute the n end-effector force data into the gravity compensation model, and calculate the sensor zero-point force and the end-effector gravity in the end-effector coordinate system; When the robotic arm interacts with the external environment, the actual interaction force is calculated based on the zero-point force of the sensor and the gravity of the end tool in the end coordinate system. The motion trajectory of the active arm is planned to obtain the state of the active arm; The state of the active boom is input into the impedance control model to generate the desired force; the state of the driven boom is adjusted according to the desired force. The active arm and the driven arm are controlled to coordinate their movements based on the states of the active arm and the driven arm.
[0007] In a preferred embodiment of the present invention, the step of constructing a gravity compensation model when the robotic arm has no interaction with the external environment includes: The gravity of the end load plus the zero-point force of the sensor is used as the first variable; The total force measured by the force sensor when the robotic arm does not interact with the external environment is used as the second variable, and the first variable is equal to the second variable to obtain the gravity compensation model.
[0008] In a preferred embodiment of the present invention, substituting N end force data into the gravity compensation model includes: Using N end force data points as the dependent variables of the gravity compensation model, the gravity compensation model is transformed into its first variant form: ; in, F 1 represents the total force of the robotic arm in its first pose without interaction with sensors. F 2 represents the total force of the robotic arm in the second pose without interaction with sensors. F n This represents the total force exerted by the robotic arm without interaction with sensors in the nth pose. I 3×3 This represents a 3×3 identity matrix. This represents the inverse of the rotation transformation matrix of the robotic arm in its first pose. This represents the inverse matrix of the rotation transformation matrix of the robotic arm in the second pose. This represents the inverse matrix of the rotation transformation matrix of the robotic arm in the nth pose; F 0 indicates zero-point force of the sensor. G b This represents the end-effector gravity in the end-effector coordinate system.
[0009] In a preferred embodiment of the present invention, calculating the sensor zero-point force and the end-effector gravity in the end-effector coordinate system includes: Transform the first variant into the second variant: ; in, f=[ F 1, F 2,..., F n ] T , f This represents the column vector of the total force of the robotic arm without interactive sensors in n poses; i =[ F 0, G b ] T , i A column vector representing the sensor zero-point force and the end-effector gravity in the end-effector coordinate system; R =[ , ,..., ] T , R The column vector of the inverse matrix of the rotation transformation matrix of the robotic arm in n poses; Using the least squares method on the second variant, the zero-point force of the sensor and the gravity of the end-effector in the end-effector coordinate system are calculated: ; in, RT express R The transpose of the matrix, express RT and R The inverse matrix of the product.
[0010] In a preferred embodiment of the present invention, the step of calculating the actual interaction force based on the zero-point force of the sensor and the gravity of the end-effector in the end-effector coordinate system when the robotic arm interacts with the external environment includes: Multiply the gravity of the end-effector in the end-effector coordinate system by the inverse of the rotation transformation matrix in the corresponding robot arm pose to obtain the gravity of the end-effector load. The zero-point force of the sensor is added to the gravity of the end load to obtain the coupling force; the coupling force is used for gravity compensation. Obtain the total force measured by the force sensor when the robotic arm interacts with the external environment under the corresponding robotic arm pose; The actual interactive force is obtained by subtracting the coupling force from the total force of the interactive sensors.
[0011] In a preferred embodiment of the present invention, before planning the motion trajectory of the active arm, the method further includes: The trajectory planning equation for the active arm is established using the following formula: ; ; in, This represents the desired acceleration of the active arm. This represents the desired velocity of the active arm. p 0 indicates the current position of the active arm. g Indicates the target position of the active arm; α This indicates the first trajectory adjustment parameter. β This indicates the second trajectory adjustment parameter. Represents a nonlinear forcing function; This represents the weight of the i-th basis function. Let i represent the i-th basis function. M This represents the total number of basis functions.
[0012] In a preferred embodiment of the present invention, the step of planning the motion trajectory of the active arm to obtain the state of the active arm includes: Based on the current position and target position of the active arm, the trajectory planning equation of the active arm is inverted: ; in, This indicates the current velocity of the active arm. p Indicates the desired position of the active arm. Indicates a time interval; The desired acceleration, desired velocity, and desired position of the active arm are taken as the active arm state.
[0013] In a preferred embodiment of the present invention, generating the desired force from the active arm state input impedance control model includes: The desired force is calculated using the following formula based on the input impedance control model of the active arm state: ; in, The desired force at time t is represented by the force. M d Represents the virtual inertia matrix. B d Represents the virtual damping matrix. K d Represents the virtual stiffness matrix; express t The expected acceleration of the active arm at any given moment. express t The expected velocity of the active arm at any given moment. express t The desired position of the active arm at any given moment.
[0014] In a preferred embodiment of the present invention, adjusting the driven boom state according to the desired force includes: The equation relating the driven boom state and the desired force is established using the following formula: ; in, Let represent the expected acceleration of the driven arm at time t. Let represent the expected velocity of the driven boom at time t. This represents the desired position of the driven boom at time t; The current acceleration, current velocity, and current position of the driven arm are collected using a robotic arm impedance controller. Adjust the current acceleration, current velocity, and current position of the driven arm until they conform to the equation between the driven arm state and the desired force.
[0015] In a preferred embodiment of the present invention, controlling the coordinated movement of the active arm and the driven arm according to the state of the active arm and the state of the driven arm includes: During the handling process, the actual internal forces are calculated based on the actual interaction forces of the driving arm and the driven arm. Subtracting the actual internal force from the expected internal force yields the internal force error; If the internal force error is greater than 0, then the actual internal force during the handling process is increased; If the internal force error is less than 0, then the actual internal force during the handling process is reduced.
[0016] The beneficial effects of this invention are as follows: The adaptive force adjustment method for a robotic arm based on gravity compensation and impedance control provided by this invention includes constructing a gravity compensation model when the robotic arm has no interaction with the external environment. In this case, the total force measured by the force sensor without interaction equals the sensor zero-point force plus the end-effector gravity in the end-effector coordinate system. N end-effector force data points are collected at different poses of the robotic arm. These n end-effector force data points are substituted into the gravity compensation model to calculate the sensor zero-point force and the end-effector gravity in the end-effector coordinate system. This eliminates the deviation caused by different robotic arms in different poses, improving the accuracy and stability of solving for the sensor zero-point force and the end-effector gravity in the end-effector coordinate system. When the robotic arm interacts with the external environment, the actual interaction force is calculated based on the sensor zero-point force and the end-effector gravity in the end-effector coordinate system. By introducing a gravity compensation mechanism, the influence of the end-effector gravity (e.g., gripper) and the sensor zero-point force under the current robotic arm pose is effectively eliminated, improving the accuracy of interaction force acquisition and solving the problem of decreased force control accuracy due to gravity interference in existing methods. The motion trajectory of the active arm is planned to obtain the active arm state. The active arm's state is input into the impedance control model to generate the desired force; the driven arm's state is then adjusted based on this desired force. By constructing an adaptive force adjustment method based on interactive states, the desired force is dynamically generated online. This allows the robotic arm's impedance controller to adjust the force output in real time based on force changes and motion states during the robotic arm's contact with the external environment, avoiding the problems of excessive or insufficient contact force in traditional fixed desired force control, and significantly improving the stability and compliance of the robotic arm's force control. The active and driven arms are controlled to coordinate their movements based on their states. In dual-robotic arm collaborative applications, this invention employs a master-slave collaborative control strategy, enabling the driven arm to respond based on the interactive force states of the active arm, achieving stable following without independent trajectory planning. This reduces the complexity of system control and improves the consistency and coordination of dual-robotic arm collaborative control. Attached Figure Description
[0017] Figure 1 This is a flowchart of the adaptive force adjustment method for a robotic arm based on gravity compensation and impedance control according to the present invention. Figure 2 This is a scene diagram of the force acquisition process at the end of the robotic arm according to the present invention; Figure 3 This is a schematic diagram of the robotic arm impedance controller of the present invention; Figure 4 This is a scene diagram of the dual-arm handling box of the present invention; Figure 5 This is a scene diagram of the invention involving two arms carrying a sphere. Detailed Implementation
[0018] Preferred embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While preferred embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the invention will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
[0019] Example 1 like Figure 1 As shown, this embodiment provides an adaptive force adjustment method for a robotic arm based on gravity compensation and impedance control, including: S1: Construct a gravity compensation model when the robotic arm does not interact with the external environment; S2: Collect n end-effector force data of the robotic arm in different poses, substitute the n end-effector force data into the gravity compensation model, and calculate the sensor zero-point force and the end-effector gravity in the end-effector coordinate system; S3: When the robotic arm interacts with the external environment, calculate the actual interaction force based on the zero-point force of the sensor and the gravity of the end tool in the end coordinate system; S4: Plan the motion trajectory of the active arm to obtain the state of the active arm; S5: Input the state of the active arm into the impedance control model to generate the desired force; adjust the state of the driven arm according to the desired force; S6: Control the coordinated movement of the active arm and the driven arm according to the state of the active arm and the state of the driven arm.
[0020] The robotic arm of this invention includes an active arm and a driven arm, both of which are equipped with end-effectors such as grippers. A six-dimensional force sensor is mounted on both the end-effector and the end of the robotic arm. The data collected by the six-dimensional force sensor consists of three parts: the force information of the six-dimensional force sensor under no-load conditions, i.e., the zero-point force of the sensor. F 0 Gravity of the end load G T And the interaction force between the robotic arm and the external environment F I .
[0021] The construction of a gravity compensation model when the robotic arm has no interaction with the external environment includes: S2: The gravity of the end load plus the zero-point force of the sensor is used as the first variable; S3: The total force measured by the force sensor when the robotic arm does not interact with the external environment is used as the second variable, and the first variable is equal to the second variable to obtain the gravity compensation model.
[0022] When the robotic arm does not interact with the external environment, the total force measured by the six-dimensional force sensor is the force of the non-interactive sensor.F Determined by the following formula: ; (1) in, F0 Indicates the zero-point force of the sensor. GT This represents the gravity of the end load. As the first variable, F As the second variable, formula (1) is the gravity compensation model.
[0023] The gravity of the end effector in the end effector coordinate system varies depending on the robot arm's pose, but the gravity of the end effector in the robot arm's base coordinate system remains constant; that is, the gravity of the end effector in the robot arm's base coordinate system is a constant. Given the rotation transformation matrices of the end effector coordinate system and the base coordinate system... and the gravity of the end tool in the end coordinate system Gb The following formula holds true: ; (2) in, It represents the inverse matrix of the rotation transformation matrix.
[0024] Substituting formula (2) into formula (1), we get: ; (3) End-effector force data acquisition and processing: A six-dimensional force sensor installed at the end of the robotic arm collects force data in real time during the contact process of the robotic arm. Combined with the current posture information of the robotic arm, the force data is uniformly converted to the task space coordinate system and fused with the gravity-compensated data to obtain force information that accurately reflects the interaction state between the end of the robotic arm and the environment, providing input basis for adaptive force adjustment.
[0025] The step of substituting N end force data into the gravity compensation model includes: Using N end force data points as the dependent variables of the gravity compensation model, the gravity compensation model is transformed into its first variant form: ; (4) in, F 1 represents the total force of the robotic arm in its first pose without interaction with sensors. F 2 represents the total force of the robotic arm in the second pose without interaction with sensors. F n This represents the total force exerted by the robotic arm without interaction with sensors in the nth pose. I 3×3 This represents a 3×3 identity matrix. This represents the inverse of the rotation transformation matrix of the robotic arm in its first pose. This represents the inverse matrix of the rotation transformation matrix of the robotic arm in the second pose. This represents the inverse matrix of the rotation transformation matrix of the robotic arm in the nth pose; F 0 indicates zero-point force of the sensor. G b This represents the end-effector gravity in the end-effector coordinate system.
[0026] The first variant, i.e., formula (4), can be simplified to formula (5): ; (5) Formula (5) is the second variant. f =[ F 1, F 2,..., F n ] T , f This represents the column vector of the total force of the robotic arm without interactive sensors in n poses; i =[ F 0, G b ] T , i A column vector representing the sensor zero-point force and the end-effector gravity in the end-effector coordinate system; R =[ , ,..., ] T , R This represents the column vector of the inverse matrix of the rotation transformation matrix of the robotic arm in n poses.
[0027] Using the least squares method on the second variant, the zero-point force of the sensor and the gravity of the end-effector in the end-effector coordinate system are calculated: ; (6) in, RT express R The transpose of the matrix, express RT and R The inverse matrix of the product.
[0028] Using formulas (4)-(6), the zero-point force of the sensor and the gravity of the end tool in the end coordinate system can be obtained, thereby realizing gravity compensation for the end tool of the robotic arm.
[0029] Under conditions where the robotic arm does not interact with the outside world, 30 different robotic arm poses are selected, and the force data collected by the six-dimensional force sensor at the end effector is measured and analyzed to achieve gravity compensation of the robotic arm's end effector. The scenario diagram of the force acquisition process at the robotic arm's end effector is shown below. Figure 2 As shown.
[0030] When the robotic arm interacts with the external environment, the actual interaction force is calculated based on the zero-point force of the sensor and the gravity of the end-effector in the end-effector coordinate system, including: S5: Multiply the gravity of the end-effector in the end-effector coordinate system by the inverse of the rotation transformation matrix in the corresponding robot arm pose to obtain the gravity of the end-effector load. Add the zero-point force of the sensor to the gravity of the end-effector load to obtain the coupling force; the coupling force is used for gravity compensation. S3: Obtain the total force measured by the force sensor when the robotic arm interacts with the external environment under the corresponding robotic arm pose; S4: Subtract the coupling force from the total force of the interactive sensors to obtain the actual interactive force.
[0031] Using formulas (4)-(6), calculate i =[ F 0, G b ] T Afterwards, G b Multiply by the inverse of the rotation transformation matrix corresponding to the robot arm's pose to obtain the gravity of the end-effector load when the robot arm interacts with the external environment.
[0032] Zero-point force of the sensor F Adding the gravity of the end-effector load when the robotic arm interacts with the external environment to obtain the coupling force, we get the coupling force. Measure the values of the six-dimensional force sensors when the robotic arm interacts with the external environment in the corresponding robotic arm pose; this is the total force from the interacting sensors. Subtract the coupling force from the total force from the interacting sensors to perform gravity compensation for the robotic arm, obtaining the true interaction force. The true interaction force is the force generated during the interaction between the robotic arm and external objects.
[0033] The adaptive force adjustment method for a robotic arm based on gravity compensation and impedance control provided in this embodiment includes constructing a gravity compensation model when the robotic arm has no interaction with the external environment. In this case, the total force measured by the force sensor without interaction equals the sensor zero-point force plus the end-effector gravity in the end-effector coordinate system. N end-effector force data points are collected at different poses of the robotic arm. These n end-effector force data points are substituted into the gravity compensation model to calculate the sensor zero-point force and the end-effector gravity in the end-effector coordinate system. This eliminates the deviation caused by different robotic arms in different poses, improving the accuracy and stability of solving for the sensor zero-point force and the end-effector gravity in the end-effector coordinate system. When the robotic arm interacts with the external environment, the actual interaction force is calculated based on the sensor zero-point force and the end-effector gravity in the end-effector coordinate system. By introducing a gravity compensation mechanism, the influence of the end-effector gravity (e.g., gripper) and the sensor zero-point force under the current robotic arm pose is effectively eliminated, improving the accuracy of interaction force acquisition and solving the problem of decreased force control accuracy due to gravity interference in existing methods. The motion trajectory of the active arm is planned to obtain the active arm state. The active arm's state is input into the impedance control model to generate the desired force; the driven arm's state is then adjusted based on this desired force. By constructing an adaptive force adjustment method based on interactive states, the desired force is dynamically generated online. This allows the robotic arm's impedance controller to adjust the force output in real time based on force changes and motion states during the robotic arm's contact with the external environment, avoiding the problems of excessive or insufficient contact force in traditional fixed desired force control, and significantly improving the stability and compliance of the robotic arm's force control. The active and driven arms are controlled to coordinate their movements based on their states. In dual-robotic arm collaborative applications, this invention employs a master-slave collaborative control strategy, enabling the driven arm to respond based on the interactive force states of the active arm, achieving stable following without independent trajectory planning. This reduces the complexity of system control and improves the consistency and coordination of dual-robotic arm collaborative control.
[0034] Example 2 This embodiment further describes the scheme of Embodiment 1 based on Embodiment 1. Before planning the motion trajectory of the active arm, the following steps are also included: To ensure the smoothness of the robotic arm's motion trajectory, this invention uses a dynamic primitive method to plan the robotic arm's speed and acceleration. Based on the planned robotic arm speed and acceleration, an impedance controller is used to adaptively adjust the desired force at the robotic arm's end effector.
[0035] DMP (Dynamic Movement Primitives) is a second-order dynamic system with spring damping, excited by nonlinear forcing terms. The trajectory planning equations of the active arm are established using DMP: ; (7) ; (8) in, This represents the desired acceleration of the active arm. This represents the desired velocity of the active arm. p 0 indicates the current position of the active arm. g Indicates the target position of the active arm; α This indicates the first trajectory adjustment parameter. β This indicates the second trajectory adjustment parameter. Represents a nonlinear forcing function; This represents the weight of the i-th basis function. Let i represent the i-th basis function. M This represents the total number of basis functions.
[0036] Represented by weighted summation of Gaussian functions The nonlinear forcing function is used to guide the trajectory of the active arm. The basis functions are used to approximate the nonlinear part of the active arm trajectory, and are expressed as follows: ; (9) Where exp represents an exponential function with the natural constant as the base. c i Indicates the first i The center of each basis function h i Indicates the first i The width of each basis function s This represents the phase of the basis functions. c i The decision was made i The basis functions in phase s Position on the axis h i Control the first i The basis functions are at the center c i The local area of effect nearby h i The larger, the more i The narrower the basis functions.
[0037] The role of the nonlinear forcing function is to enable the robotic arm not only to converge to the target point, but also to accurately reproduce the shape of the taught trajectory, and to generalize this shape to different starting points, target points, and time scales. The role of robotic arm trajectory planning is to automatically generate a motion sequence from the initial pose to the target pose, which is continuous in time, under the premise of satisfying motion constraints and task requirements.
[0038] The process of planning the motion trajectory of the active arm to obtain the state of the active arm includes: S4: Based on the current position and target position of the active arm, inverse the trajectory planning equation of the active arm: ; (10) in, This indicates the current velocity of the active arm. p Indicates the desired position of the active arm. Indicates a time interval; S5: The desired acceleration, desired velocity, and desired position of the active arm are taken as the active arm state.
[0039] DMP based on the current position of the active arm p 0 and the target position of the active arm g Planning the trajectory of the robotic arm involves considering the following factors: acceleration determines the relationship between motion and force, as well as the smoothness of the trajectory; velocity determines how fast the robotic arm moves along the trajectory; and the desired position determines the overall orientation of the robotic arm's motion.
[0040] The step of generating the desired force from the input impedance control model of the active arm state includes: The desired force is calculated using the following formula based on the input impedance control model of the active arm state: ; (11) in, express t The expected force at that moment, M d Represents the virtual inertia matrix. B d Represents the virtual damping matrix. K d Represents the virtual stiffness matrix; express t The expected acceleration of the active arm at any given moment. express t The expected velocity of the active arm at any given moment. express t The desired position of the active arm at any given moment.
[0041] The adjustment of the driven boom state according to the desired force includes: The equation relating the driven boom state and the desired force is established using the following formula: ;(12) in, Let represent the expected acceleration of the driven arm at time t. Let represent the expected velocity of the driven boom at time t. This represents the desired position of the driven boom at time t; S6: Use a robotic arm impedance controller to collect the current acceleration, current velocity, and current position of the driven arm; adjust the current acceleration, current velocity, and current position of the driven arm until they conform to the equation between the driven arm state and the desired force.
[0042] Impedance-controlled dual-arm collaboration includes: (1) sending instructions to the active arm of the dual-arm robot system; the DMP plans the trajectory by combining the current position and target position of the active arm; and calculates the desired acceleration, desired velocity, and desired position of the active arm according to formula (10). (2) inputting the state of the active arm into the impedance control model, and calculating the desired position according to formula (11). t The expected force at the moment is adjusted so that the expected force of the active arm and the expected force of the driven arm are equal. When the active arm and the driven arm jointly clamp or operate a rigid object, if the expected forces of the two are equal in magnitude, opposite in direction and collinear, a force closure can be formed inside the object, so as not to generate net force / torque and avoid the active arm and the driven arm pushing and pulling each other. (3) Based on the impedance controller of the robotic arm, the current acceleration, current velocity and current position of the driven arm autonomous movement are obtained. Under the action of impedance control, the posture of the driven arm is adjusted according to the expected force to achieve smooth contact between the active arm and the driven arm, and provide sufficient tracking speed and acceleration for the driven arm, so as to prevent the object from slipping during the coordinated action of the two arms, such as the carrying of the object.
[0043] Figure 3 This is the schematic diagram of the impedance controller for the robotic arm. Figure 3 x in r For the reference position trajectory, k is the stiffness coefficient, and f r Let b be the desired contact force, b be the damping coefficient, and x be the actual position. e For position error, k e Here is the end stiffness matrix.
[0044] Example 3 This embodiment further describes the scheme of Embodiment 1 based on Embodiments 1 and 2. The step of controlling the coordinated movement of the active arm and the driven arm according to the states of the active arm and the driven arm includes: S2: During the handling process, the actual internal force is calculated based on the actual interaction force of the driving arm and the actual interaction force of the driven arm; The internal force error is obtained by subtracting the actual internal force from the expected internal force. If the internal force error is greater than 0, the actual internal force during the handling process is increased; if the internal force error is less than 0, the actual internal force during the handling process is decreased.
[0045] The actual interaction force of the active arm after gravity compensation and the actual interaction force of the driven arm after gravity compensation were measured according to the method in Example 1. When the actual interaction forces of the active arm and the driven arm are in opposite directions, the actual interaction force of the active arm is subtracted from the actual interaction force of the driven arm to obtain the interaction force difference. The interaction force difference is divided by 2 to obtain the actual internal force.
[0046] The internal force error is obtained by subtracting the actual internal force from the expected internal force (the force exerted by the mutual compression between the driving and driven arms). The internal force error characterizes the deviation of the internal force of the two arms in gripping the object; the smaller the internal force error, the stronger the stability of the two arms in grasping or gripping the object.
[0047] If the internal force error is greater than 0, it means that the expected internal force is greater than the actual internal force. In this case, increase the actual internal force during the handling process, add the internal force error to the expected force of the drive arm, and subtract the internal force error from the expected force of the driven arm to reduce the internal force error. If the internal force error is less than 0, it means that the expected internal force is less than the actual internal force. In this case, decrease the actual internal force during the handling process, subtract the internal force error from the expected force of the drive arm, and add the internal force error to the expected force of the driven arm to reduce the internal force error.
[0048] This embodiment adjusts the desired forces of the active arm and the driven arm through an internal force constraint mechanism. The active arm and the driven arm complete the object handling process through continuous force information interaction and adaptive adjustment, while maintaining overall force balance and system stability. This effectively suppresses the internal force conflict and uneven force distribution that may occur during the dual-arm collaboration process, and improves the stability and safety of the system during handling and contact operations.
[0049] Figure 4 This is a scene illustration of the dual-arm transport box of the present invention. Figure 5 This is a scene illustration of the invention involving two arms transporting a sphere. Figures 4-5 As shown, after completing gravity compensation, DMP-based trajectory generation, and impedance parameter adjustment, the target object's pose is first acquired, and the gripping points and end effector postures of the dual arms are planned. Subsequently, based on the DMP trajectory control, the dual arms synchronously approach the target. During the contact phase, compliant contact is achieved through impedance control, and adaptive force compensation is performed by combining end effector force sensing information to achieve coordinated force balance and stable transport. Finally, the gripping force is gradually released at the target position to achieve stable placement of the object, completing the dual-arm transport task and returning to the initial state.
[0050] The method provided by this invention can achieve stable and reliable force control under different load conditions and different contact stiffness environments. It is applicable to various contact operation scenarios such as handling and assembly, without the need for complex models or hardware modifications, and has high engineering practicality and promotion value.
[0051] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.
[0052] The above description is only a preferred embodiment of this application and does not limit the patent scope of this application. Any equivalent structural or procedural changes made based on the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.
Claims
1. A method for adaptive force adjustment of a robot arm based on gravity compensation and impedance control, characterized in that, include: When the robotic arm does not interact with the external environment, a gravity compensation model is constructed; Collect n end-effector force data of the robotic arm in different poses, substitute the n end-effector force data into the gravity compensation model, and calculate the sensor zero-point force and the end-effector gravity in the end-effector coordinate system; When the robotic arm interacts with the external environment, the actual interaction force is calculated based on the zero-point force of the sensor and the gravity of the end tool in the end coordinate system. The motion trajectory of the active arm is planned to obtain the state of the active arm; The desired force is generated by using the input impedance control model of the active arm state. Adjust the driven boom state according to the desired force; The active arm and the driven arm are controlled to coordinate their movements based on the states of the active arm and the driven arm.
2. The method of claim 1, wherein, The construction of a gravity compensation model when the robotic arm has no interaction with the external environment includes: The gravity of the end load plus the zero-point force of the sensor is used as the first variable; The total force measured by the force sensor when the robotic arm does not interact with the external environment is used as the second variable, and the first variable is equal to the second variable to obtain the gravity compensation model.
3. The method of claim 2, wherein the method further comprises: The step of substituting N end force data into the gravity compensation model includes: Using N end force data points as the dependent variables of the gravity compensation model, the gravity compensation model is transformed into its first variant form: ; in, F 1 represents the total force of the robotic arm in its first pose without interaction with sensors. F 2 represents the total force of the robotic arm in the second pose without interaction with sensors. F n This represents the total force exerted by the robotic arm without interaction with sensors in the nth pose. I 3×3 This represents a 3×3 identity matrix. This represents the inverse of the rotation transformation matrix of the robotic arm in its first pose. This represents the inverse matrix of the rotation transformation matrix of the robotic arm in the second pose. This represents the inverse matrix of the rotation transformation matrix of the robotic arm in the nth pose; F 0 indicates zero-point force of the sensor. G b This represents the end-effector gravity in the end-effector coordinate system.
4. The adaptive force adjustment method for a robotic arm based on gravity compensation and impedance control according to claim 3, characterized in that, The calculation of the sensor zero-point force and the end-effector gravity in the end-effector coordinate system includes: Transform the first variant into the second variant: ; in, f =[ F 1, F 2,..., F n ] T , f This represents the column vector of the total force of the robotic arm without interactive sensors in n poses; i =[ F 0, G b ] T , i A column vector representing the sensor zero-point force and the end-effector gravity in the end-effector coordinate system; R =[ , ,..., ] T , R The column vector of the inverse matrix of the rotation transformation matrix of the robotic arm in n poses; Using the least squares method on the second variant, the zero-point force of the sensor and the gravity of the end-effector in the end-effector coordinate system are calculated: ; in, RT express R The transpose of the matrix, express RT and R The inverse matrix of the product.
5. The adaptive force adjustment method for a robotic arm based on gravity compensation and impedance control according to claim 3, characterized in that, When the robotic arm interacts with the external environment, the actual interaction force is calculated based on the zero-point force of the sensor and the gravity of the end-effector in the end-effector coordinate system, including: Multiply the gravity of the end-effector in the end-effector coordinate system by the inverse of the rotation transformation matrix in the corresponding robot arm pose to obtain the gravity of the end-effector load. The zero-point force of the sensor is added to the gravity of the end load to obtain the coupling force; the coupling force is used for gravity compensation. Obtain the total force measured by the force sensor when the robotic arm interacts with the external environment under the corresponding robotic arm pose; The actual interactive force is obtained by subtracting the coupling force from the total force of the interactive sensors.
6. The adaptive force adjustment method for a robotic arm based on gravity compensation and impedance control according to claim 1, characterized in that, Before planning the motion trajectory of the active arm, the process also includes: The trajectory planning equation for the active arm is established using the following formula: ; ; in, This represents the desired acceleration of the active arm. This represents the desired velocity of the active arm. p 0 indicates the current position of the active arm. g Indicates the target position of the active arm; α This indicates the first trajectory adjustment parameter. β This indicates the second trajectory adjustment parameter. Represents a nonlinear forcing function; This represents the weight of the i-th basis function. Let i represent the i-th basis function. M This represents the total number of basis functions.
7. The adaptive force adjustment method for a robotic arm based on gravity compensation and impedance control according to claim 6, characterized in that, The process of planning the motion trajectory of the active arm to obtain the state of the active arm includes: Based on the current position and target position of the active arm, the trajectory planning equation of the active arm is inverted: ; in, This indicates the current velocity of the active arm. p Indicates the desired position of the active arm. Indicates a time interval; The desired acceleration, desired velocity, and desired position of the active arm are taken as the active arm state.
8. The adaptive force adjustment method for a robotic arm based on gravity compensation and impedance control according to claim 7, characterized in that, The step of generating the desired force from the input impedance control model of the active arm state includes: The desired force is calculated using the following formula based on the input impedance control model of the active arm state: ; in, The desired force at time t is represented by the force. M d Represents the virtual inertia matrix. B d Represents the virtual damping matrix. K d Represents the virtual stiffness matrix; express t The expected acceleration of the active arm at any given moment. express t The expected velocity of the active arm at any given moment. express t The desired position of the active arm at any given moment.
9. The adaptive force adjustment method for a robotic arm based on gravity compensation and impedance control according to claim 8, characterized in that, The adjustment of the driven boom state according to the desired force includes: The equation relating the driven boom state and the desired force is established using the following formula: ; in, Let represent the expected acceleration of the driven arm at time t. Let represent the expected velocity of the driven boom at time t. This represents the desired position of the driven boom at time t; The current acceleration, current velocity, and current position of the driven arm are collected using a robotic arm impedance controller. Adjust the current acceleration, current velocity, and current position of the driven arm until they conform to the equation between the driven arm state and the desired force.
10. The adaptive force adjustment method for a robotic arm based on gravity compensation and impedance control according to claim 1, characterized in that, The step of controlling the coordinated movement of the active arm and the driven arm according to the states of the active arm and the driven arm includes: During the handling process, the actual internal forces are calculated based on the actual interaction forces of the driving arm and the driven arm. Subtracting the actual internal force from the expected internal force yields the internal force error; If the internal force error is greater than 0, then the actual internal force during the handling process is increased; If the internal force error is less than 0, then the actual internal force during the handling process is reduced.
