A dexterous hand grasping method, system, medium, and device based on adaptive impedance optimization

Abstract

This invention discloses a dexterous hand grasping method, system, medium, and device based on adaptive impedance optimization. It includes acquiring contact information to construct a contact state space model, determining whether to perform impedance parameter optimization and executing optimization or parameter maintenance, calculating and correcting a reference control force, and controlling the dexterous hand based on the corrected safe control force. Through NMPC-based impedance parameter optimization, the impedance parameters are adjusted online according to changes in the contact state, improving the dexterous hand's environmental adaptability during the grasping process of unknown objects. Furthermore, by dividing the grasping process into states and combining preset optimization trigger conditions to trigger impedance parameter optimization on demand, continuous high-frequency optimization throughout the process is avoided, reducing the real-time computational burden of the control algorithm. Finally, a safety filter equation based on a control barrier function is introduced to correct the reference control input online, effectively ensuring that the contact force remains within a safe range and reducing the risk of object damage and grasping instability.

Description

Technical Field

[0001] This application relates to the field of dexterous hand control technology, and in particular to a dexterous hand grasping method, system, medium and device based on adaptive impedance optimization. Background Technology

[0002] When a dexterous hand grasps an unknown object, it not only needs to establish a stable grip but also needs to control the contact force to ensure it does not exceed the upper limit of the object's tolerance, thus avoiding irreversible structural damage or plastic deformation to flexible or low-yield-strength objects. Therefore, unknown object grasping control usually needs to simultaneously consider grasping stability, interaction compliance, safety, and real-time performance.

[0003] In existing dexterous hand grasping control, the most common method is fixed-parameter impedance control. This type of method, by pre-setting impedance stiffness and damping parameters, gives the dexterous hand a certain degree of compliance during contact, and the control structure is relatively simple and easy to implement. However, its impedance parameters are usually determined empirically, making it difficult to simultaneously adapt to the grasping needs of different types of objects with high, medium, and low stiffness. When the impedance parameter is set too high, it is easy to generate large contact impacts and peak forces; when the impedance parameter is set too low, it may lead to insufficient support capacity, unstable grasping, or slow response.

[0004] To address the challenge of real-time matching of fixed impedance parameters to changes in the contact environment, existing technologies typically employ adaptive impedance control to adjust impedance parameters online. Some methods update impedance parameters according to preset rules by detecting changes in contact force, displacement, velocity, or the contact environment; however, these methods have limited adaptability under complex conditions. Other methods utilize learning-based impedance control strategies, training a learning model using historical interaction data to establish a mapping relationship between contact state and impedance parameters, achieving data-driven impedance adjustment. However, these methods rely on data quality and model generalization ability, and their stability and safety remain insufficient in unknown scenarios.

[0005] Building upon this, some methods employ impedance parameter optimization based on nonlinear model predictive control (NMPC). These methods combine contact dynamics models to comprehensively consider factors such as indentation depth, contact force, and impedance parameter variations within the prediction time domain, and can explicitly handle contact force constraints, displacement constraints, and parameter boundary constraints. Compared to impedance adjustment methods that rely solely on empirical parameter tuning or local rule corrections, NMPC-based impedance parameter optimization can uniformly address multi-objective and multi-constraint problems in the grasping process, making it more suitable for online impedance adjustment during the grasping of unknown objects. However, continuously solving nonlinear optimization problems at high frequency throughout the grasping process, especially in scenarios involving multiple fingers and multiple contact points of a dexterous hand, will impose a significant online computational burden, impacting the real-time response capability and engineering implementation efficiency of the control system.

[0006] Meanwhile, existing grasping control methods are often not clear enough in terms of grasping state division and impedance parameter update timing, lacking a mechanism to autonomously determine "when to optimize and when to maintain" based on different stages such as contact establishment, characteristic changes, and stable grasping. Furthermore, the handling of contact force safety boundaries often relies on empirical thresholds, simple saturation limits, or offline tuning, lacking a strict online safety constraint mechanism. When model errors, sensor noise, or external disturbances exist, problems such as contact force exceeding the limit, object damage, or grasping instability may still occur.

[0007] Therefore, how to ensure that the contact force remains within a safe range while triggering impedance parameter optimization only when necessary during the grasping of unknown objects has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0008] Therefore, it is necessary to provide a dexterous hand grasping method, system, medium, and device based on adaptive impedance optimization to address the above-mentioned technical problems.

[0009] The following technical solution is adopted in this specification: This manual provides a dexterous hand grasping method based on adaptive impedance optimization, specifically including: The system acquires contact measurement information between the fingertip and the target object during the current control cycle. It extracts the current indentation depth and normal indentation velocity in the normal direction of the contact point between the fingertip and the object from the contact measurement information to form the current contact state. Based on the equivalent dynamic relationship of the fingertip-object contact, it establishes a contact state evolution model for impedance control, which includes undetermined parameters for impedance control. The undetermined parameters for impedance control include at least impedance stiffness and impedance damping.

[0010] Based on the current contact state and historical contact state information, it is determined whether the preset optimization triggering conditions are met. If so, an adaptive impedance optimization equation with constraints based on nonlinear model predictive control (NMPC) is constructed according to the current contact state. The contact state evolution model is used as the prediction model, and the current indentation depth and normal indentation velocity are used as inputs to calculate the current optimal impedance parameters. Otherwise, the historical impedance parameters from the previous control cycle are directly used as the current optimal impedance parameters. The constraints include at least the normal contact force in the prediction time domain not being greater than the maximum safe contact force that the object can withstand, the feasible region constraint of the indentation depth, and the boundary constraint of the impedance parameters.

[0011] Based on the current contact state and optimal impedance parameters, a reference control force is calculated, a control barrier function is introduced, and a safety filter equation is constructed based on the control barrier function to correct the reference control force, thereby obtaining the safety control force.

[0012] The dexterous hand is controlled according to the aforementioned safety control force.

[0013] Furthermore, the equations for the contact state evolution model, which includes undetermined parameters for impedance control, are as follows: ; in, For compression acceleration, For the depth of indentation, For the pressing speed, For the desired push depth, For impedance stiffness, For impedance damping, For equivalent stiffness, For equivalent damping, For contact equivalent mass.

[0014] Furthermore, the preset optimization triggering conditions include at least contact establishment or contact loss, contact state change exceeding a set threshold, or contact force change trend indicating a decrease in grasping stability.

[0015] Furthermore, the determination of whether the preset optimized triggering conditions are met based on the current contact state and historical contact state information is implemented by a finite state machine; the finite state machine includes at least a contact detection and physical feature evaluation state, an impedance adaptive compliance adjustment state, and a stable grasping and holding state.

[0016] When the fingertip makes contact with the object being grasped, the contact establishment conditions are met, and the system enters the contact detection and physical feature evaluation state. It collects contact force, indentation depth, indentation speed, and related contact features, and estimates the dynamic parameters of the current contact environment. After completing contact feature identification, it switches to the impedance adaptive compliance adjustment state and calculates the current optimal impedance parameters according to the adaptive impedance optimization equation to achieve adaptive matching of the impedance parameters to the current contact conditions. When the impedance parameters tend to stabilize, it enters the stable grasping and holding state, where the impedance parameters remain unchanged to reduce the computational burden caused by continuous repetitive optimization.

[0017] When the preset optimized triggering condition is triggered, the contact state changes, and the system switches back to the impedance adaptive compliant adjustment state to calculate the current optimal impedance parameters until the impedance parameters tend to stabilize, at which point it enters the stable gripping and holding state.

[0018] Furthermore, the objective function of the adaptive impedance optimization equation based on nonlinear model predictive control (NMPC) is to minimize the following cost function: ; in, For the prediction time domain of NMPC, For the stage cost function, For the terminal cost function, For the first The push depth at each time step and All are weighting coefficients.

[0019] Furthermore, the security filtering equation is as follows: ; ; in, For reference control force, For actual control, The convergence coefficient is . The reaction force of the object being grasped on the fingertip. and These are the equivalent parameters for stiffness and damping, respectively. This is the maximum safe contact force that an object can withstand.

[0020] Furthermore, after controlling the dexterous hand according to the safety control force, the process also includes using the current contact state and the current optimal impedance parameter as historical contact state information to re-enter the next control cycle until the dexterous hand completes the current grasping task.

[0021] This specification provides a dexterous hand grasping system based on adaptive impedance optimization, specifically including: The contact state evolution model construction module is used to acquire contact measurement information between the fingertip and the target object in the current control cycle, extract the current indentation depth and normal indentation velocity in the normal direction of the contact point between the fingertip and the object to form the current contact state, and establish a contact state evolution model for impedance control.

[0022] The optimization judgment and execution module is used to determine whether the preset optimization trigger conditions are met based on the current contact state and historical contact state information, and to perform pending parameter optimization or directly use the historical impedance parameters of the previous control cycle as the current optimal impedance parameters based on the judgment result.

[0023] The control force calculation and correction module is used to calculate the reference control force and correct the reference control force by constructing a safety filtering equation to obtain the safety control force.

[0024] The dexterous hand control module is used to control the dexterous hand according to the safety control force.

[0025] This specification provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described above.

[0026] This specification provides a computer device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method described above.

[0027] The beneficial effects of this invention are as follows: Compared with the prior art, this invention improves the environmental adaptability of the dexterous hand in grasping unknown objects by optimizing impedance parameters based on NMPC and adjusting the impedance parameters online according to changes in contact state. Furthermore, by dividing the grasping process into states and combining preset optimization trigger conditions to trigger impedance parameter optimization on demand, this invention avoids continuous high-frequency optimization throughout the entire process, reducing the real-time computational burden of the control algorithm. Finally, by introducing a safety filter equation based on the control obstacle function, the reference control input is corrected online, effectively ensuring that the contact force is always within a safe range and reducing the risk of object damage and grasping instability. Attached Figure Description

[0028] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0029] Figure 1 This is a flowchart of the adaptive impedance and safety interactive capture control in an embodiment of the present invention. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this specification clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments in this specification without creative effort are within the scope of protection of this application.

[0031] This invention provides a dexterous hand grasping method based on adaptive impedance optimization, comprising four steps: collecting contact information to construct a contact state space model; determining whether to perform impedance parameter optimization and executing optimization or parameter maintenance; calculating and correcting the reference control force; and controlling the dexterous hand based on the corrected safety control force. Through NMPC-based impedance parameter optimization, the impedance parameters are adjusted online according to changes in the contact state, improving the dexterous hand's environmental adaptability during the grasping process of unknown objects. Furthermore, by dividing the grasping process into states and combining preset optimization trigger conditions to trigger impedance parameter optimization on demand, continuous high-frequency optimization throughout the process is avoided, reducing the real-time computational burden of the control algorithm. Finally, a safety filtering equation based on a control barrier function is introduced to correct the reference control input online, effectively ensuring that the contact force remains within a safe range and reducing the risk of object damage and grasping instability.

[0032] The technical solutions provided by the various embodiments of this application are described in detail below with reference to the accompanying drawings.

[0033] A dexterous hand grasping method based on adaptive impedance optimization specifically includes the following steps: Step 1: Obtain contact measurement information between the fingertip and the target object in the current control cycle. Extract the current indentation depth and normal indentation velocity in the normal direction of the contact point between the fingertip and the object from the contact measurement information to form the current contact state. Based on the equivalent dynamic relationship of the fingertip-object contact, establish a contact state evolution model for impedance control, which includes undetermined parameters for impedance control. The undetermined parameters for impedance control include at least impedance stiffness and impedance damping.

[0034] Specifically, the contact state evolution model for impedance control, which includes undetermined parameters for impedance control, is as follows: ; in, For compression acceleration, For the depth of indentation, For the pressing speed, For the desired push depth, For impedance stiffness, For impedance damping, For equivalent stiffness, For equivalent damping, For contact equivalent mass.

[0035] Step 2: Based on the current contact state and historical contact state information, determine whether the preset optimization triggering conditions are met. If so, construct a constrained adaptive impedance optimization equation based on nonlinear model predictive control (NMPC) according to the current contact state, using the contact state evolution model as the prediction model, and the current indentation depth and normal indentation velocity as inputs, to calculate the current optimal impedance parameters; otherwise, directly use the historical impedance parameters from the previous control cycle as the current optimal impedance parameters. The constraints include at least the normal contact force in the prediction time domain not exceeding the maximum safe contact force that the object can withstand, the feasible region constraint of the indentation depth, and the boundary constraints of the impedance parameters.

[0036] Specifically, the preset optimization trigger conditions include at least contact establishment or contact loss, contact state change exceeding a set threshold, or contact force change trend indicating decreased grasping stability.

[0037] Furthermore, based on the current contact state and historical contact state information, it is determined whether the preset optimized triggering conditions are met through a finite state machine; the finite state machine includes at least: contact detection and physical feature evaluation state, impedance adaptive compliance adjustment state, and stable grasping and holding state.

[0038] When the fingertip makes contact with the object being grasped, the contact establishment conditions are met, and the system enters the contact detection and physical feature evaluation state. It collects contact force, indentation depth, indentation speed, and related contact features, and estimates the dynamic parameters of the current contact environment. After completing contact feature identification, it switches to the impedance adaptive compliance adjustment state and calculates the current optimal impedance parameters according to the adaptive impedance optimization equation to achieve adaptive matching of the impedance parameters to the current contact conditions. When the impedance parameters tend to stabilize, it enters the stable grasping and holding state, where the impedance parameters remain unchanged to reduce the computational burden caused by continuous repetitive optimization. When the preset optimized trigger condition is triggered, the contact state changes, and it switches back to the impedance adaptive compliant adjustment state to calculate the current optimal impedance parameter until the impedance parameter tends to stabilize, and then enters the stable grasp and hold state.

[0039] Preferably, the objective function of the adaptive impedance optimization equation for the NMPC based on nonlinear model predictive control is to minimize the following cost function: ; in, For the prediction time domain of NMPC, For the stage cost function, For the terminal cost function, For the first The push depth at each time step and All are weighting coefficients.

[0040] Step 3: Calculate the reference control force based on the current contact state and optimal impedance parameters, introduce the control barrier function, construct a safety filter equation based on the control barrier function to correct the reference control force, and obtain the safety control force.

[0041] Specifically, the security filtering equation is: ; ; in, For reference control force, For actual control, The convergence coefficient is . The reaction force of the object being grasped on the fingertip. and These are the equivalent parameters for stiffness and damping, respectively. This is the maximum safe contact force that an object can withstand.

[0042] Step 4: Control the dexterous hand according to the safety control force.

[0043] Preferably, after controlling the dexterous hand according to the safety control force, the control cycle also includes using the current contact state and the current optimal impedance parameter as historical contact state information to re-enter the next control cycle until the dexterous hand completes the grasping task.

[0044] Example This embodiment comprehensively considers contact state evolution, impedance parameter optimization, event-triggered switching, and safety constraint control. For example... Figure 1 As shown, the specific steps include the following:

[0045] S1: Collect contact information and then construct a contact state space model. In the normal direction of the contact point, the local contact process between the nimble fingertip and the object is equivalent to an impedance control object.

[0046] Let the state variables be: ; in, For the depth of indentation, The indentation velocity. Based on the environmental equivalent stiffness. Environmental equivalent damping and contact equivalent quality Establish the state-space equations:

[0047] make , ; ; in, and These are the stiffness coefficient and damping coefficient of the impedance control to be optimized, respectively. This indicates the desired indentation depth. A reference normal contact force is generated based on the impedance relationship:

[0048] ; in, The desired injection speed.

[0049] S2: Considering that the contact dynamics parameters may change with the contact state during the grasping of an unknown object, and that the impedance parameter adjustment needs to simultaneously satisfy multiple constraints such as indentation depth, contact force, and impedance parameter boundaries, based on the state-space model of step 1, and combined with the contact dynamics parameters... In other words, stiffness and damping are considered, and an adaptive impedance optimization model based on NMPC is constructed. When the preset optimization triggering condition is met, the current moment can be obtained by solving the following optimal control problem. The optimal impedance parameters are as follows:

[0050] ; in, This represents the prediction time domain of NMPC; As the initial state of the current optimization problem, it is constrained to be equal to the real-time measurement value of the sensor. To ensure the stability of the closed-loop control; Represents the discrete dynamic model of the system; the state at the next moment. From the current state and control input Decision; furthermore, to ensure the safety of the grasping process, the optimal control problem also needs to satisfy the condition that the fingertip force of the dexterous hand cannot exceed a threshold. And the indentation depth is less than Meanwhile, the impedance coefficient obtained through iteration must be within a reasonable range; furthermore, Represents the stage cost function. The terminal cost function is expressed as follows: ; The stage cost consists of three parts: the first term represents the deviation cost of the contact penetration depth; the second term limits the contact force amplitude to avoid instability or damage caused by excessive force; and the third term is a penalty for the increment of impedance parameter changes to ensure the smoothness of the control process. Among these, , For the corresponding weighting coefficients, This is the corresponding weight matrix; the terminal cost mainly consists of two terms: the first term represents the convergence error of the final indentation depth; the second term represents the indentation speed error, ensuring that the finger and object are relatively stationary during stable grasping. , These are the corresponding weighting coefficients.

[0051] By solving the above NMPC optimization problem, the impedance stiffness can be adjusted at each sampling time based on the current contact state and contact dynamic parameters. and impedance damping Online optimization is performed to achieve adaptive and compliant adjustment of the grasping process for unknown objects.

[0052] S3: Considering that continuous online solving by NMPC will bring a large computational burden, an event triggering mechanism based on FSM is designed for the entire process of dexterous hand grasping.

[0053] The entire grasping process is divided into three consecutive control sub-states: contact detection and physical feature evaluation state, impedance adaptive compliance adjustment state, and stable grasping and holding state.

[0054] When a finger comes into contact with the object being grasped, the system first enters the contact detection and physical feature evaluation state, collecting contact force, indentation depth, indentation speed and related contact features, and estimating the dynamic parameters of the current contact environment.

[0055] After completing the contact feature identification, the system switches to the impedance adaptive compliance adjustment state, and optimizes the impedance parameters online using the NMPC model in step 2 to achieve adaptive matching of the impedance parameters to the current contact conditions.

[0056] Once the impedance parameters stabilize, the system enters a stable hold state, where the impedance parameters remain unchanged to reduce the computational burden caused by continuous repeated optimization.

[0057] When the contact state changes again, the impedance parameter optimization process is triggered again. The changes in the contact state mainly include: displacement of the object, increase in contact force, and significant changes in contact dynamic parameters.

[0058] Through the aforementioned FSM and event-triggered mechanism, the relationship between NMPC's global optimization capabilities and the system's real-time computational burden can be effectively balanced, enabling the dexterous hand to maintain grasping compliance and stability while reducing the computational pressure caused by continuous high-frequency optimization.

[0059] S4: To ensure security during the grabbing process, a CBF-based security filtering mechanism is further introduced. The barrier function is defined as follows:

[0060] ; in, This is the maximum safe contact force that the object can withstand. For the current normal contact force, For the depth of indentation, The pressing speed.

[0061] According to Nagumo's theorem, to ensure system security, the following conditions must be met: , And at the point of contact: ; in For contact equivalent mass, The actual control force applied to the fingertips, The contact reaction force generated by the object, its value is... equal.

[0062] Then, the reference control input is transformed into a quadratic programming (QP) problem with safety constraints for solution, namely: ; ; in, For reference control force, For actual control, The convergence coefficient is . The reaction force of the object being grasped on the fingertip. and These are the equivalent parameters for stiffness and damping, respectively. This is the maximum safe contact force that an object can withstand.

[0063] S5: Map the safety-corrected control quantity to the dexterous hand execution instruction and complete the dexterous hand grasping action.

[0064] Will As a safety control input for the dexterous hand's end contact point, and combined with the current configuration of the dexterous hand and the normal direction of the contact point, it is mapped to the execution commands of each finger joint, driving the dexterous hand to complete the grasping of the target object. During the grasping execution, if the current contact force, indentation depth, and indentation speed meet the corresponding safety constraints, the current control output is maintained to achieve stable grasping; if the contact state changes again, steps 3 and 4 are re-triggered to re-optimize the impedance parameters and correct the control input online to ensure the safety and adaptability of the grasping process.

[0065] This specification provides a dexterous hand grasping system based on adaptive impedance optimization, including: The contact state evolution model construction module is used to acquire contact measurement information between the fingertip and the target object in the current control cycle, extract the current indentation depth and normal indentation velocity in the normal direction of the contact point between the fingertip and the object to form the current contact state, and establish a contact state evolution model for impedance control.

[0066] The optimization judgment and execution module is used to determine whether the preset optimization trigger conditions are met based on the current contact state and historical contact state information, and to perform pending parameter optimization or directly use the historical impedance parameters of the previous control cycle as the current optimal impedance parameters based on the judgment result.

[0067] The control force calculation and correction module is used to calculate the reference control force and correct it by constructing a safety filtering equation to obtain the safety control force.

[0068] The dexterous hand control module is used to control the dexterous hand according to the safety control force.

[0069] Specific limitations regarding the dexterous hand grasping system based on adaptive impedance optimization can be found in the limitations of the dexterous hand grasping method based on adaptive impedance optimization described above, and will not be repeated here. Each module in the aforementioned dexterous hand grasping system based on adaptive impedance optimization can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in hardware or independently of the processor in a computer device, or stored in software in the memory of a computer device, so that the processor can call and execute the corresponding operations of each module.

[0070] This specification also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0071] This specification provides a computer device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method described above.

[0072] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical storage, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0073] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A dexterous hand grasping method based on adaptive impedance optimization, characterized in that, Specifically, it includes: The system acquires contact measurement information between the fingertip and the target object during the current control cycle. It then extracts the current indentation depth and normal indentation velocity at the contact point between the fingertip and the object, constituting the current contact state. Based on the equivalent dynamic relationship of the fingertip-object contact, it establishes a contact state evolution model for impedance control, including undetermined parameters for impedance control. These undetermined parameters include at least impedance stiffness and impedance damping. Based on the current contact state and historical contact state information, it is determined whether the preset optimization triggering conditions are met. If so, an adaptive impedance optimization equation with constraints based on nonlinear model predictive control (NMPC) is constructed according to the current contact state. The contact state evolution model is used as the prediction model, and the current indentation depth and normal indentation velocity are used as inputs to calculate the current optimal impedance parameters. Otherwise, the historical impedance parameters of the previous control cycle are directly used as the current optimal impedance parameters. The constraints include at least the normal contact force in the prediction time domain not being greater than the maximum safe contact force that the object can withstand, the feasible region constraint of the indentation depth, and the boundary constraint of the impedance parameters. Based on the current contact state and optimal impedance parameters, a reference control force is calculated, a control barrier function is introduced, and a safety filter equation is constructed based on the control barrier function to correct the reference control force, thereby obtaining the safety control force. The dexterous hand is controlled according to the aforementioned safety control force.

2. The dexterous hand grasping method based on adaptive impedance optimization as described in claim 1, characterized in that, The equations for the contact state evolution model, which includes undetermined parameters for impedance control, are as follows: ； in, For compression acceleration, For the depth of indentation, For the pressing speed, For the desired push depth, For impedance stiffness, For impedance damping, For equivalent stiffness, For equivalent damping, For contact equivalent mass.

3. The dexterous hand grasping method based on adaptive impedance optimization as described in claim 1, characterized in that: The preset optimization trigger conditions include at least contact establishment or contact loss, contact state change exceeding a set threshold, or contact force change trend indicating decreased grasping stability.

4. The dexterous hand grasping method based on adaptive impedance optimization as described in claim 1, characterized in that: The determination of whether the preset optimized triggering conditions are met based on the current contact state and historical contact state information is implemented by a finite state machine; the finite state machine includes at least a contact detection and physical feature evaluation state, an impedance adaptive compliance adjustment state, and a stable grasping and holding state. When the fingertip makes contact with the object being grasped, the contact establishment conditions are met, and the system enters the contact detection and physical feature evaluation state. This state collects contact force, indentation depth, indentation speed, and related contact features, and estimates the dynamic parameters of the current contact environment. After completing contact feature identification, the system switches to the impedance adaptive compliance adjustment state and calculates the current optimal impedance parameters according to the adaptive impedance optimization equation to achieve adaptive matching of the impedance parameters to the current contact conditions. Once the impedance parameters stabilize, the system enters the stable grasping and holding state, maintaining the impedance parameters unchanged in this state to reduce the computational burden caused by continuous repetitive optimization. When the preset optimized triggering condition is triggered, the contact state changes, and the system switches back to the impedance adaptive compliant adjustment state to calculate the current optimal impedance parameters until the impedance parameters tend to stabilize, at which point it enters the stable gripping and holding state.

5. The dexterous hand grasping method based on adaptive impedance optimization as described in claim 1, characterized in that, The objective function of the adaptive impedance optimization equation based on nonlinear model predictive control (NMPC) is to minimize the following equivalence function: ； in, For the prediction time domain of NMPC, For the stage cost function, For the terminal cost function, For the first The push depth at each time step and All are weighting coefficients.

6. The dexterous hand grasping method based on adaptive impedance optimization as described in claim 1, characterized in that, The security filtering equation is as follows: ； ； in, For reference control force, For actual control, The convergence coefficient is . The reaction force of the object being grasped on the fingertip. and These are the equivalent parameters for stiffness and damping, respectively. This is the maximum safe contact force that an object can withstand.

7. The dexterous hand grasping method based on adaptive impedance optimization as described in claim 1, characterized in that: After controlling the dexterous hand according to the safety control force, the process further includes using the current contact state and the current optimal impedance parameter as historical contact state information to re-enter the next control cycle until the dexterous hand completes the current grasping task.

8. A dexterous hand grasping system based on adaptive impedance optimization, characterized in that, include: The contact state evolution model construction module is used to acquire contact measurement information between the dexterous fingertip and the target object in the current control cycle, extract the current indentation depth and normal indentation velocity in the normal direction of the contact point between the fingertip and the object to form the current contact state, and establish a contact state evolution model for impedance control. The optimization judgment and execution module is used to determine whether the preset optimization trigger conditions are met based on the current contact state and historical contact state information, and to perform pending parameter optimization or directly use the historical impedance parameters of the previous control cycle as the current optimal impedance parameters based on the judgment result. The control force calculation and correction module is used to calculate the reference control force and correct the reference control force by constructing a safety filtering equation to obtain the safety control force; The dexterous hand control module is used to control the dexterous hand according to the safety control force.

9. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the method described in any one of claims 1 to 7.

10. A computer device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method described in any one of claims 1 to 7.

Images