A variable stiffness joint external force estimation method and device

Abstract

The application provides a variable stiffness joint external force estimation method and device, comprising: acquiring a stiffness value in real time; substituting the acquired stiffness value into an adaptive fusion coefficient model to solve and obtain an adaptive fusion coefficient; obtaining external force estimation I based on current information of the variable stiffness joint; obtaining external force estimation II based on elastic deformation information and the stiffness value of the variable stiffness joint; and obtaining external force estimation based on the adaptive fusion coefficient, the external force estimation I and the external force estimation II, and fusing the current and the elastic deformation information. The application fuses the current and the elastic deformation information in the external force estimation process, improves the external force estimation effect under dynamic and quasi-static stiffness, and is suitable for a variable stiffness joint system requiring high-precision external force estimation in a force interaction scene.

Description

Technical Field

[0001] This invention belongs to the field of robotics technology, specifically relating to a method and apparatus for estimating external forces in a variable stiffness joint. Background Technology

[0002] Variable stiffness joints, with their compliant properties, are widely used in scenarios where robots interact with their surroundings. In complex interactive tasks, high-precision force sensing is an important future development direction. Traditional force sensing is mainly achieved through external force sensors; however, external force sensors are expensive and their application is limited by constraints such as complex structures and additional installation space.

[0003] In recent years, external force sensing methods other than force sensors have been developed. Chinese Patent 201410112717.5 discloses a method based on H... ∞ A filtered method for estimating the external force of a robotic arm is proposed, deriving a dynamic model without explicit acceleration. The external force is treated as a state variable in the state-space equation, based on H... ∞ Filtering is used to estimate external forces. Chinese Patent 202210279718.3 discloses a method for estimating tactile external forces of a robotic arm based on a mechanistic data hybrid model. It estimates external forces based on current information and uses a supervised statistical learning method of Gaussian process regression to train compensation terms for unmodeled residual dynamic models based on datasets, thereby compensating for unknown disturbances. Chinese Patent 202210207511.5 discloses a method and system for estimating robot external forces based on configuration Jacobi condition number optimization. It uses a generalized momentum observer to estimate the external forces acting on the robot and improves the accuracy of external force estimation by accurately modeling frictional forces. Configuration Jacobi condition number optimization is used to filter the robot configuration, suppressing the influence of joint torque noise and modeling errors on external force estimation.

[0004] However, force estimation based on current information requires an accurate dynamic model. For variable stiffness joints, which use a dual-drive configuration of position motor and stiffness adjustment motor, the complexity of the dynamic model and the unmodeled residual parts restrict the effectiveness of force estimation.

[0005] Since variable stiffness joints have internal compliance properties, external forces can be estimated by detecting the elastic deformation inside the joint. However, this method has a better external force estimation effect under quasi-static stiffness, but its external force estimation effect is often poor under dynamic stiffness.

[0006] Furthermore, for lever-type variable stiffness joints, the maximum allowable external torque of the joint is different under different stiffnesses. Ignoring this factor will lead to saturation output of the detection torque.

[0007] Therefore, how to achieve high-precision external force estimation for variable stiffness joints is an urgent problem to be solved. Summary of the Invention

[0008] To address the shortcomings of existing technologies, this invention provides a method and apparatus for estimating external forces in variable stiffness joints, which can achieve high-precision external force estimation in variable stiffness joint systems.

[0009] The present invention achieves the above-mentioned technical objectives through the following technical means.

[0010] The first aspect of this invention provides a method for estimating external forces in a variable stiffness joint, comprising the following steps:

[0011] Real-time acquisition of stiffness values;

[0012] The obtained stiffness values ​​are substituted into the adaptive fusion coefficient model for calculation to obtain the adaptive fusion coefficients; the adaptive fusion coefficient model is as follows:

[0013]

[0014] in, R represents the absolute value of the derivative of joint stiffness K. w1 and R w2 All are positive integers, satisfying R w2 >R w1 ;

[0015] External force estimation I is obtained based on the current information of the variable stiffness joint;

[0016] External force estimation based on elastic deformation information and stiffness values ​​of variable stiffness joints II;

[0017] Based on the adaptive fusion coefficient, external force estimation I, and external force estimation II, the external force estimate based on the fusion of current and elastic deformation information is obtained:

[0018] τ q =K τ τ ei +(1-K τ )τ ec

[0019] Where, τ ei For external force estimation I, τ ec For external force estimation II, K τ For adaptive fusion coefficients.

[0020] In one embodiment of the first aspect, the step of acquiring the stiffness value in real time is as follows:

[0021] Design a stiffness zone controller and calculate the expected value of stiffness variables based on the estimated external force and maximum external moment constraint at the previous moment.

[0022] The expected value of the stiffness variable is applied to the motion control of the stiffness-adjusting motor to obtain the actual value of the stiffness variable;

[0023] Substituting the actual values ​​of the stiffness variables into the stiffness model yields the updated stiffness values, where the stiffness model is a function of the stiffness variables.

[0024] In one embodiment of the first aspect, when the variable stiffness joint is a lever-type variable stiffness joint, the design stiffness region controller calculates the expected value of the stiffness variable based on the estimated external force and the maximum external moment constraint at the previous moment, including:

[0025] (1) Design a stiffness region controller based on the Lyapunov barrier function;

[0026] For lever-type variable stiffness joints, the Lyapunov barrier function used for stiffness region control is expressed as:

[0027]

[0028] Where η represents the function variable, η=|τ q | / τ em -η0; η0, η1, and η2 are all positive constants and satisfy η0 = η2 = 0.5, η1 < 0.5; τ q τ represents the estimated external torque. em This represents the maximum allowable external moment of the joint under a specific stiffness variable γ, and is related to the stiffness variable γ; |τ q | / τ em The value range is [0,1];

[0029] (2) Based on the Lyapunov barrier function, stiffness adjustment is divided into three regions: stiffness increase region, stiffness decrease region and constant stiffness region.

[0030] (3) Based on the region involved by the Lyapunov barrier function, its derivative is expressed as:

[0031]

[0032] (4) Design a suitable Make it satisfy To ensure that η is in the constant stiffness region, The expression is:

[0033]

[0034] Where, k η kη is a constant representing how quickly the stiffness of a lever-type variable stiffness joint transitions from a region of increasing stiffness or decreasing stiffness to a region of constant stiffness, and kη is greater than 0.

[0035] (5) According to the definition of η, the derivative of η can be expressed as:

[0036]

[0037] in, The derivative of the stiffness variable γ of the lever-type variable stiffness joint;

[0038] (6) After further sorting, the following results were obtained The expression:

[0039]

[0040] In one embodiment of the first aspect, the acquisition of external force estimation I based on the current information of the variable stiffness joint includes:

[0041] A dynamic model of the position drive module and the stiffness adjustment module is established. The output torque of the position drive module and the stiffness adjustment module is obtained by combining the measured joint current value. The sum of the two is used as the external force estimate I.

[0042] The dynamic models of the position drive module and the stiffness adjustment module are expressed as follows:

[0043]

[0044] in, and These represent the moments of inertia of the position drive module and the stiffness adjustment module, respectively. and These represent the frictional forces of the position drive module and the stiffness adjustment module, respectively. and These represent the output torques of the position drive module and the stiffness adjustment module, respectively. and θ1 and θ2 represent the input torques of the position drive module and stiffness adjustment module, respectively; θ1 and θ2 represent the output angular displacements of the position drive module and stiffness adjustment module, respectively.

[0045] After establishing the dynamic model, the moment of inertia in the dynamic model is obtained through the dynamic parameter identification method. and friction The accurate value.

[0046] Then, the current information I of the position motor and the stiffness adjustment motor is measured. p and I s The input torques of the position drive module and the stiffness adjustment module are calculated according to the following formula:

[0047]

[0048] Among them, K piand K si These represent the total torque constants of the position drive module and the stiffness adjustment module, respectively.

[0049] After obtaining the input torques of the position drive module and the stiffness adjustment module, the input torques of the position drive module and the stiffness adjustment module are substituted into their respective dynamic models for calculation to obtain the output torques of the position drive module and the stiffness adjustment module. and

[0050] Calculate the output torque of the position drive module and the stiffness adjustment module. and The sum of these is used as an estimate of the external force I, i.e.

[0051] In one embodiment of the first aspect, the expression for the external force estimation II is:

[0052]

[0053] Where K(γ) represents the stiffness value, This indicates the elastic deformation inside the joint, which is measured using additional sensors.

[0054] A second aspect of the present invention provides a lever-type variable stiffness joint external force estimation device, comprising:

[0055] The stiffness acquisition module is configured to acquire stiffness values ​​in real time.

[0056] The external force estimation I acquisition module is configured to acquire external force estimation I based on the current information of the variable stiffness joint;

[0057] The External Force Estimation II acquisition module is configured to acquire External Force Estimation II based on the elastic deformation information and stiffness values ​​of the variable stiffness joint;

[0058] An adaptive fusion coefficient model is configured to obtain updated stiffness values ​​based on the actual values ​​of stiffness variables, wherein the adaptive fusion coefficient model is:

[0059]

[0060] in, R represents the absolute value of the derivative of joint stiffness K. w1 and R w2 All are positive integers, satisfying R w2 >R w1 ;

[0061] The fusion module is used to obtain an external force estimate based on the fusion of current and elastic deformation information, using adaptive fusion coefficients, external force estimation I, and external force estimation II.

[0062] τ q =K τ τ ei +(1-K τ )τ ec ;

[0063] Where, τ ei For external force estimation I, τ ec For external force estimation II, K τ For adaptive fusion coefficients.

[0064] In one embodiment of the second aspect, the stiffness acquisition module includes:

[0065] The stiffness variable expectation value calculation module is configured as a design stiffness zone controller, which calculates the expectation value of stiffness variables based on the estimated external force value and the maximum external moment constraint at the previous moment.

[0066] The stiffness variable actual value acquisition module is configured to apply the expected value of the stiffness variable to the motion control of the stiffness adjustment motor to obtain the actual value of the stiffness variable;

[0067] The stiffness model is configured to obtain updated stiffness values ​​based on the actual values ​​of stiffness variables, where the stiffness model is a function of the stiffness variables.

[0068] In one embodiment of the second aspect, the external force estimation I acquisition module includes:

[0069] The dynamics model building module is configured to create dynamics models for the position-driven module and the stiffness adjustment module:

[0070]

[0071] in, and These represent the moments of inertia of the position drive module and the stiffness adjustment module, respectively. and These represent the frictional forces of the position drive module and the stiffness adjustment module, respectively. and These represent the output torques of the position drive module and the stiffness adjustment module, respectively. and θ1 and θ2 represent the input torques of the position drive module and stiffness adjustment module, respectively; θ1 and θ2 represent the output angular displacements of the position drive module and stiffness adjustment module, respectively.

[0072] The parameter identification module is configured to obtain the moment of inertia in the dynamic model through a dynamic parameter identification method. and friction The accurate value;

[0073] The external force calculation module is configured to calculate the current information I based on the measured position motor and stiffness adjustment motor. p and I s The input torques of the position drive module and the stiffness adjustment module are calculated using the following formula:

[0074]

[0075] Among them, K pi and K si These represent the total torque constants of the position drive module and the stiffness adjustment module, respectively;

[0076] The output torque calculation module is configured to substitute the calculated input torques of the position drive module and stiffness adjustment module into the dynamic models of the position drive module and stiffness adjustment module for solution, thereby obtaining the output torques of the position drive module and stiffness adjustment module; and to calculate the sum of the output torques of the position drive module and stiffness adjustment module as the external force estimate I.

[0077]

[0078] Compared with the prior art, this invention has outstanding substantive features and significant progress, specifically:

[0079] The external force estimation method proposed in this invention achieves the fusion of external force estimation based on current and external force estimation based on elastic deformation information through fusion coefficients, which makes up for the shortcomings of single external force estimation methods, improves the external force estimation effect under dynamic and quasi-static stiffness, and is suitable for variable stiffness joint systems that require high-precision external force estimation in force interaction scenarios.

[0080] This invention designs an adaptive fusion coefficient model based on stiffness variables. The adaptive fusion coefficient changes autonomously with the stiffness value, which can realize a smooth transition between the external force estimation value based on current and the external force estimation value based on elastic deformation, and avoid the jump of the external force estimation value.

[0081] This invention designs a stiffness region controller that uses the estimated external force value and the maximum external moment constraint from the previous moment to achieve dynamic updates of stiffness values, and the stiffness values ​​are closer to the actual stiffness values.

[0082] Based on the updated stiffness variables and the measured elastic deformation inside the joint, this invention derives an external force estimate II based on elastic deformation information, thereby improving the external force estimation effect under dynamic stiffness.

[0083] In estimating external forces based on current information, this invention comprehensively considers the dynamic information of the position drive module and the stiffness adjustment module, obtains the output torque of the position drive module and the stiffness adjustment module respectively, and uses the sum of the two as the overall output torque of the joint, making the overall output torque of the joint closer to the actual output torque and improving the accuracy of external force estimation based on current information. The dynamic model of each sub-drive also includes unknown parameters that can be identified through dynamic parameters and measurable angular displacements, further reducing the impact of model errors on external force estimation.

[0084] This invention takes into account that the maximum allowable external torque of the lever-type variable stiffness joint varies under different stiffnesses. It designs a stiffness region controller, which dynamically adjusts the stiffness variable to keep the stiffness in the constant stiffness region, thus ensuring that the joint has good external force estimation resolution and low stiffness. Attached Figure Description

[0085] Figure 1 This is a flowchart of the variable stiffness joint external force estimation method in Example 1;

[0086] Figure 2 This is a schematic diagram of the adaptive fusion coefficients;

[0087] Figure 3 This is a flowchart of the variable stiffness joint external force estimation method in Example 2;

[0088] Figure 4 This is a flowchart of the variable stiffness joint external force estimation method in Example 3;

[0089] Figure 5 This represents the correspondence between the maximum allowable external torque and the stiffness variable of a lever-type variable stiffness joint.

[0090] Figure 6 A simplified diagram of stiffness region control;

[0091] Figure 7 A comparison of the results of three different external force estimation methods.

[0092] Figure 8 This is a schematic diagram of the structure of Example 4. Detailed Implementation

[0093] Embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein similar or identical reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0094] Variable stiffness joints are a crucial component in realizing flexible robots, enabling robots to achieve interactive tasks. In multi-scenario, multi-type force interaction tasks, high-precision external force perception is an important future development direction.

[0095] Currently, mainstream external force sensing methods include force estimation methods based on current information and those based on elastic deformation. For current-based force estimation, the common dual-drive structure of variable stiffness joints (using a position motor and a stiffness motor to adjust joint position and stiffness, respectively) leads to complex dynamic models. This complexity, along with unmodeled dynamic components, limits the effectiveness of current-based force estimation. Specifically, current-based force estimation shows stable performance under both dynamic and quasi-static stiffness conditions, but its performance under quasi-static stiffness is inferior to that based on elastic deformation. As for elastic deformation-based force estimation, its performance is better under quasi-static stiffness conditions due to resolution limitations, but worse under dynamic stiffness conditions.

[0096] To address the aforementioned issues, this invention presents a fusion external force estimation method and apparatus that integrates external force estimation based on current and external force estimation based on elastic deformation information. This overcomes the shortcomings of single external force estimation methods, improves the external force estimation performance under dynamic and quasi-static stiffness conditions, and is suitable for variable stiffness joint systems requiring high-precision external force estimation in force interaction scenarios.

[0097] The following will discuss exemplary embodiments of the invention in detail with reference to the accompanying drawings.

[0098] Example 1

[0099] like Figure 1 As shown in the figure, this embodiment provides a method for estimating the external force of a variable stiffness joint. The specific steps are as follows:

[0100] Step S1: Obtain stiffness values ​​in real time.

[0101] Step S2, Design the adaptive fusion coefficient model:

[0102]

[0103] Where K represents joint stiffness. R represents the absolute value of the derivative of joint stiffness K. w1 and R w2 All are positive integers, satisfying R w2 >R w1 ;

[0104] Substituting the updated stiffness values ​​into the adaptive fusion coefficient model, we obtain the adaptive fusion coefficient K.τ .

[0105] Step S3: Obtain the external force estimate Iτ based on the current information of the variable stiffness joint. ei .

[0106] Step S4: Obtain the external force estimate IIτ based on the elastic deformation information of the variable stiffness joint and the updated stiffness value. ec :

[0107]

[0108] Where K is the stiffness value of the joint. This indicates the elastic deformation inside the joint, which is measured using additional sensors.

[0109] It can be seen that, in calculating the external force estimation II based on elastic deformation information, this embodiment not only uses the measured elastic deformation inside the joint. It also incorporates a dynamic stiffness value K, thereby improving the estimation effect of external forces under dynamic stiffness.

[0110] Step S5, based on the adaptive fusion coefficient K τ External force estimation Iτ ei and external force estimation IIτ ec This yields an estimate of the external force based on the fusion of current and elastic deformation information.

[0111] τ q =K τ τ ei +(1-K τ )τ ec .

[0112] In some embodiments, K τ It can also be a preset fusion coefficient designed based on the relationship between current information and elastic deformation information.

[0113] like Figure 2 As shown, the adaptive fusion coefficient K calculated according to step S2 τ The diagram illustrates the adaptive fusion coefficient K. τ According to Self-regulation, when At that time, K τ The value is 1, and the external force estimation based on current information plays a dominant role; when At that time, K τ When the value is 0, the external force estimation based on elastic deformation information plays a dominant role; when At that time, the estimated value of external force was calculated by combining the current and elastic deformation.

[0114] By adapting the fusion coefficient K τ The design of the expression in the form of a sine curve can achieve a smooth transition between the external force estimation value based on current and the external force estimation value based on elastic deformation, avoiding jumps in the external force estimation value.

[0115] In some embodiments, the variable stiffness joint is a lever-type variable stiffness joint or a cam-type variable stiffness joint, etc. That is, the external force estimation method proposed in this embodiment can be applied to lever-type variable stiffness joint systems, as well as cam-type variable stiffness joint systems, etc.

[0116] Example 2

[0117] The difference between this embodiment and embodiment 1 is that: an embodiment of step S3 is given, such as... Figure 3 As shown.

[0118] The specific steps are as follows:

[0119] Step S31: Establish the dynamic models of the position drive module and the stiffness adjustment module.

[0120] The dynamic model is expressed as follows:

[0121]

[0122] in, and These represent the moments of inertia of the position drive module and the stiffness adjustment module, respectively, and are unknown parameters; and These represent the frictional forces of the position drive module and the stiffness adjustment module, respectively, and are unknown parameters; and These represent the output torques of the position drive module and the stiffness adjustment module, respectively. and θ1 and θ2 represent the input torques of the position drive module and stiffness adjustment module, respectively; θ1 and θ2 represent the output angular displacements of the position drive module and stiffness adjustment module, respectively.

[0123] Step S32: Obtain the moment of inertia in the dynamic model using the dynamic parameter identification method. and friction The accurate value.

[0124] As can be seen from step S31, the dynamic model includes both motion parameters θ1 and θ2, as well as unknown dynamic parameters. and Motion parameters θ1 and θ2 can be obtained through measurement, but for dynamic parameters that cannot be measured... and Then it is necessary to obtain the accurate values ​​of the parameters through dynamic parameter identification methods.

[0125] Dynamic parameter identification, in particular, is the process of solving for unknown parameters of the joint dynamic model by combining the dynamic parameter model with the angular position error of the joint output end using appropriate mathematical methods. Many algorithms can be used for dynamic parameter identification, such as maximum likelihood estimation, least squares method, particle swarm optimization, Kalman filtering, neural networks, and genetic algorithms.

[0126] It should be noted that in a variable stiffness joint, the motion parameters include not only the position drive module angle θ1 and the stiffness adjustment module angle θ2, but also the internal elastic deformation. And the joint output angular displacement q, while satisfying the following constraints: These motion parameters can all be acquired using additional sensors.

[0127] Step S33: Measure the motor current values ​​of the position drive module and the stiffness adjustment module, and calculate the input torque of the position drive module and the stiffness adjustment module according to the following formula:

[0128]

[0129] Among them, K pi and K si These represent the total torque constants of the position drive module and the stiffness adjustment module, respectively, reflecting parameters such as the motor torque constant and reduction ratio; I p and I s These represent the motor current values ​​for the position drive module and the stiffness adjustment module, respectively.

[0130] Step S34: Substitute the calculated input torques of the position drive module and stiffness adjustment module into the dynamic model of the position drive module and stiffness adjustment module for solution, and obtain the output torques of the position drive module and stiffness adjustment module. and

[0131] Step S35: The output torque of the position drive module and the stiffness adjustment module is... and The operation yields an external force estimate I based on current information:

[0132] As can be seen, in this embodiment, when estimating external forces based on current information, the output torque of the stiffness adjustment module is introduced after comprehensively considering the dynamic information of the position drive module and the stiffness adjustment module. By obtaining the output torques of the position drive module and the stiffness adjustment module respectively, and using their sum as the overall output torque of the joint, the overall output torque of the joint is closer to the actual output torque, thus improving the accuracy of external force estimation based on current information. The dynamic model of each sub-drive also includes unknown parameters that can be identified through dynamic parameters and measurable angular displacements, further reducing the impact of model errors on external force estimation.

[0133] Example 3

[0134] The difference between this embodiment and Embodiment 1 is that, as Figure 4 As shown, the steps for obtaining the stiffness value in real time are as follows:

[0135] Step S11: Design a stiffness region controller based on the estimated external force τ from the previous moment. q and maximum external moment constraint τ em Calculate the expected value γ of the stiffness variable. d ;

[0136] Step S12: Apply the expected value of the stiffness variable to the stiffness adjustment motor, and obtain the actual value γ of the stiffness variable through the motion control of the stiffness adjustment motor. r ;

[0137] Step S12, change the actual value of the stiffness variable γ r Substituting into the stiffness model K(γ), we obtain the updated stiffness value K(γ). r As can be seen, the stiffness model K(γ) is a function of the stiffness variable γ.

[0138] In this embodiment of the invention, a stiffness region controller is designed, utilizing the estimated external force τ from the previous moment. q and maximum external moment constraint τ em It achieves dynamic updates of stiffness values, and the stiffness values ​​are closer to the actual stiffness values.

[0139] Example 4

[0140] This embodiment defines the variable stiffness joint as a lever-type variable stiffness joint system and provides a specific implementation method.

[0141] The specific steps are as follows:

[0142] Step A: Obtain stiffness values ​​in real time;

[0143] Step A1: Design a stiffness region controller based on the estimated external force τ from the previous moment. q and maximum external moment constraint τem The expected value of the stiffness variable γ was calculated. d .

[0144] The specific steps are as follows:

[0145] Step 1: Design a stiffness region controller based on the Lyapunov barrier function.

[0146] For lever-type variable stiffness joints, the Lyapunov barrier function used for stiffness region control is expressed as:

[0147]

[0148] Where η represents the function variable, η=|τ q | / τ em -η0; η0, η1, and η2 are all positive constants and satisfy η0 = η2 = 0.5, η1 < 0.5; τ q τ represents the estimated external torque. em This represents the maximum allowable external moment of the joint under a specific stiffness variable γ, and is related to the stiffness variable γ; |τ q | / τ em The value range is [0,1].

[0149] like Figure 5 As shown, the maximum allowable external torque τ of the joint em Constrained by both the maximum compression and rated torque of the elastic element, τ is not a constant value, but rather related to the stiffness variable γ: τ is within the range [0, γ1]. em τ increases with increasing γ, and this process is mainly constrained by the maximum compression of the elastic element; in the interval [γ1, γ2], τ em It is a constant value and is subject to the rated torque constraint.

[0150] Understandably, due to |τ q | / τ em The value range of η is [0,1]. Therefore, by introducing η0 and defining η0 = η2 = 0.5, η can be made to be between [-0.5, 0.5].

[0151] Step 2, based on the Lyapunov barrier function, divides the stiffness adjustment into three regions: a stiffness-increasing region, a stiffness-decreasing region, and a constant stiffness region; specifically as follows... Figure 6 As shown, the region division is based on the parameter η.

[0152] Specifically, when η > η1, |τ q | / τ em When the ratio is close to or greater than 1, the external torque τ acting on the joint q Approaching or exceeding the maximum withstandable moment τ at this stiffness emIncrease the control stiffness K to allow the maximum allowable torque τ em It also increases to ensure that the external forces on the joint are within the acceptable range, hence it is called the "stiffness increase zone";

[0153] When η < -η1, |τ q | / τ em When the ratio is close to 0, the external torque τ acting on the joint is... q Much smaller than the maximum allowable moment τ at this stiffness em By controlling the stiffness K to decrease, the resolution of external force estimation is improved, hence it is called the "stiffness reduction region";

[0154] When -η1 < η < η1, |τ q | / τ em Within a reasonable range, the joint has a suitable external force estimation resolution, while the torque it bears is also less than the maximum allowable torque τ. em At this point, the joint is in the "constant stiffness region" and no adjustment is made to the stiffness.

[0155] It is understandable that the constant stiffness region is the desired region, and the purpose of stiffness adjustment is to keep the stiffness within the constant stiffness region.

[0156] Step 3, based on the region involved by the Lyapunov barrier function, its derivative is expressed as:

[0157]

[0158] Step 4, design a suitable Make it satisfy To ensure that η is in the constant stiffness region, The expression is:

[0159]

[0160] Where, k η Let k be a constant representing how quickly the stiffness of a variable stiffness joint transitions from a region of increasing stiffness or decreasing stiffness to a region of constant stiffness, and k η It is greater than 0, and its value can be selected according to actual needs.

[0161] The constant stiffness region can guarantee the external torque τ q The maximum allowable moment τ under a certain stiffness em This also ensures that the joint is in a low stiffness and high external force resolution; therefore, it is necessary to keep η in the constant stiffness region as much as possible, so as to avoid the stiffness being in a constant adjustment process and save energy.

[0162] Furthermore, when the direction of the external torque changes, |τ q | / τem The ratio fluctuates around 0. By setting a minimum stiffness value, stiffness fluctuations can be avoided.

[0163] Step 5, according to the definition of η, η|=τ q | / τ em -η0 expresses the derivative of η as:

[0164]

[0165] in, The derivative of the stiffness variable γ of the variable stiffness joint reflects the magnitude of the stiffness adjustment rate and affects how fast the stiffness adjustment is.

[0166] Step 6, after further processing, obtain The expression:

[0167]

[0168] As can be seen, in this embodiment, considering the different maximum allowable external torques of the joint under different stiffnesses, a stiffness region controller was designed. By adjusting the stiffness variable, it is ensured that η is in the constant stiffness region, which guarantees that the joint has good external force estimation resolution and avoids saturation output of the detected torque.

[0169] Step A2, after calculating the expected value γ of the stiffness variable. d Then, considering the expected value γ of the stiffness variable calculated by the stiffness zone controller... d This is only a theoretical value; therefore, the expected value γ of the stiffness variable will also be considered. d Motion control applied to the stiffness-adjusting motor yields the actual value γ of the stiffness variable. r This enables real-time updates of stiffness variables.

[0170] Furthermore, when applied to discrete-time control systems, the stiffness variable γ is expressed as:

[0171]

[0172] Among them, t i This represents the i-th sampling time; Δt is the system's sampling period, t i+1 =t i +Δt.

[0173] Step A3, change the actual value of the stiffness variable γ r Substituting into the stiffness model K(γ), we obtain the updated stiffness value K(γ). r ).

[0174] Step B: Obtain the adaptive fusion coefficients.

[0175] Step B1, Design the adaptive fusion coefficient model:

[0176]

[0177] Where K represents joint stiffness. R represents the absolute value of the derivative of joint stiffness K. w1 and R w2 All are positive integers, satisfying R w2 >R w1 ;

[0178] Step B2, update the stiffness value K(γ) r Substituting into the adaptive fusion coefficient model, the adaptive fusion coefficient K is obtained. τ .

[0179] Step C, obtain external force estimation I based on the current information of the variable stiffness joint:

[0180] In practice, the following steps are included:

[0181] Step C1: Establish the dynamic models of the position drive module and the stiffness adjustment module.

[0182] The dynamic model is expressed as follows:

[0183]

[0184] in, and These represent the moments of inertia of the position drive module and the stiffness adjustment module, respectively, and are unknown parameters; and These represent the frictional forces of the position drive module and the stiffness adjustment module, respectively, and are unknown parameters; and These represent the output torques of the position drive module and the stiffness adjustment module, respectively. and θ1 and θ2 represent the input torques of the position drive module and stiffness adjustment module, respectively; θ1 and θ2 represent the output angular displacements of the position drive module and stiffness adjustment module, respectively.

[0185] Step C2: Obtain the moment of inertia in the dynamic model using the dynamic parameter identification method. and friction The accurate value;

[0186] Step C3: Measure the motor current values ​​of the position drive module and the stiffness adjustment module, and calculate the input torque of the position drive module and the stiffness adjustment module according to the following formula:

[0187]

[0188] Among them, K pi and K si These represent the total torque constants of the position drive module and the stiffness adjustment module, respectively, reflecting parameters such as the motor torque constant and reduction ratio; I p and I s These represent the motor current values ​​for the position drive module and the stiffness adjustment module, respectively.

[0189] Step C4: Substitute the calculated input torques of the position drive module and stiffness adjustment module into the dynamic models of the position drive module and stiffness adjustment module for solution, and obtain the output torques of the position drive module and stiffness adjustment module. and

[0190] Step C5: The output torque of the position drive module and the stiffness adjustment module is... and The operation yields an external force estimate I based on current information:

[0191] As can be seen, in this embodiment, the dynamic model is decoupled and split into the dynamic model of the position drive module and the dynamic model of the stiffness adjustment module, thereby reducing the high-order dynamic model to the control of each sub-drive, simplifying the model and reducing the control difficulty; the dynamic model of each sub-drive includes unknown parameters that can be identified through dynamic parameters and measurable angular displacement, further reducing the impact of model error on external force estimation.

[0192] Step D, based on the elastic deformation information of the variable stiffness joint and the updated stiffness value K(γ) r Obtaining External Force Estimation II:

[0193]

[0194] Step E, based on the adaptive fusion coefficient K τ External force estimation Iτ ei and external force estimation IIτ ec This yields an estimate of the external force based on the fusion of current and elastic deformation information.

[0195] τ q =K τ τ ei +(1-K τ )τ ec .

[0196] like Figure 7 The figure shows an experimental comparison of three different external force estimation methods in a lever-type variable stiffness joint scenario. The average absolute value of the errors of the three different external force estimation results is shown below:

[0197]

[0198] It can be seen that the external force estimation based on elastic deformation has a good external force estimation effect under quasi-static stiffness, but the external force estimation effect under dynamic stiffness is poor.

[0199] The force estimation based on current information has a stable force estimation effect under both dynamic and quasi-static stiffness conditions, but its force estimation effect under quasi-static stiffness conditions is worse than that based on elastic deformation.

[0200] The external force estimation method proposed in this invention, which integrates current and elastic deformation information, combines the advantages of both types of information, improves the external force estimation effect under dynamic and quasi-static stiffness, and has the smallest average value of the overall absolute error in terms of external force estimation error.

[0201] Example 4

[0202] This embodiment also provides a lever-type variable stiffness joint external force estimation device, such as Figure 8 As shown, it includes:

[0203] The stiffness acquisition module is configured to acquire stiffness values ​​in real time.

[0204] The external force estimation I acquisition module is configured to acquire the external force estimation Iτ based on the current information of the variable stiffness joint. ei ;

[0205] The external force estimation II acquisition module is configured to acquire elastic deformation information and updated stiffness value γ based on the variable stiffness joint. r Obtaining external force estimation IIτ ec ;

[0206] In practical implementation, the expression for the external force estimation II is:

[0207]

[0208] Where K(γ) represents the stiffness value, This indicates elastic deformation within the joint;

[0209] The self-fusion coefficient model is configured to be based on the actual value γ of the stiffness variable. r Obtain the updated stiffness value K(γ) r ), wherein the adaptive fusion coefficient model is:

[0210]

[0211] Where K represents joint stiffness, which is a function of the stiffness variable. R represents the absolute value of the derivative of joint stiffness K.w1 and R w2 All are positive integers, satisfying R w2 >R w1 ;

[0212] The fusion module is used to fused based on the adaptive fusion coefficient K. τ External force estimation Iτ ei and external force estimation IIτ ec This yields an estimate of the external force based on the fusion of current and elastic deformation information.

[0213] τ q =K τ τ ei +(1-K τ )τ ec .

[0214] It can be seen that the external force estimation device proposed in this embodiment adopts an external force estimation method that combines external force estimation based on current and external force estimation based on elastic deformation information, which makes up for the shortcomings of a single external force estimation method and improves the external force estimation effect under dynamic and quasi-static stiffness.

[0215] The adaptive fusion coefficient model changes autonomously with the stiffness variable, enabling a smooth transition between external force estimates based on current and external force estimates based on elastic deformation, thus avoiding fluctuations in the external force estimates.

[0216] The external force estimation device proposed in this embodiment not only uses the measured elastic deformation inside the joint when calculating the external force estimation II based on elastic deformation information, but also combines the stiffness value acquired in real time, which improves the external force estimation effect under dynamic stiffness.

[0217] Furthermore, the stiffness acquisition module proposed in this embodiment includes:

[0218] The stiffness variable expectation value calculation module is configured as a design stiffness region controller, based on the previous time-stress estimate τ. q and maximum external moment constraint τ em The expected value of the stiffness variable γ was calculated. d ;

[0219] The stiffness variable actual value acquisition module is configured to apply the expected value of the stiffness variable to the motion control of the stiffness adjustment motor to obtain the actual value γ of the stiffness variable. r ;

[0220] Stiffness model, configured to be based on the actual value γ of stiffness variable. r Obtain the updated stiffness value K(γ) r ), where the stiffness model is represented as K(γ), which is a function of the stiffness variable γ.

[0221] Furthermore, the external force estimation I acquisition module proposed in this embodiment includes:

[0222] The dynamics model building module is configured to create dynamics models for the position-driven module and the stiffness adjustment module:

[0223]

[0224] in, and These represent the moments of inertia of the position drive module and the stiffness adjustment module, respectively. and These represent the frictional forces of the position drive module and the stiffness adjustment module, respectively. and These represent the output torques of the position drive module and the stiffness adjustment module, respectively. and θ1 and θ2 represent the input torques of the position drive module and stiffness adjustment module, respectively; θ1 and θ2 represent the output angular displacements of the position drive module and stiffness adjustment module, respectively.

[0225] The parameter identification module is configured to obtain the moment of inertia in the dynamic model through a dynamic parameter identification method. and friction The accurate value;

[0226] The external force calculation module is configured to calculate the current information I based on the measured position motor and stiffness adjustment motor. p and I s The input torques of the position drive module and the stiffness adjustment module are calculated using the following formula:

[0227]

[0228] Among them, K pi and K si These represent the total torque constants of the position drive module and the stiffness adjustment module, respectively;

[0229] The output torque calculation module is configured to substitute the calculated input torques of the position drive module and stiffness adjustment module into the dynamic models of the position drive module and stiffness adjustment module for solution, thereby obtaining the output torques of the position drive module and stiffness adjustment module; and to calculate the sum of the output torques of the position drive module and stiffness adjustment module as the external force estimate I.

[0230]

[0231] When estimating external forces based on current information, the output torque of the stiffness adjustment module is introduced to improve the accuracy of external force estimation based on current information. Specifically, based on the comprehensive consideration of the dynamic information of the position drive module and the stiffness adjustment module, the output torques of the position drive module and the stiffness adjustment module are obtained separately, and the sum of the two is used as the overall output torque of the joint, so that the overall output torque of the joint is closer to the actual output torque. The dynamic model of each sub-drive includes unknown parameters that can be identified through dynamic parameters and measurable angular displacements, further reducing the impact of model errors on external force estimation.

[0232] While the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the scope of the invention. Any person skilled in the art can make modifications without departing from the scope of the invention; all equivalent modifications made in accordance with the invention should be covered by the scope of the invention.

Claims

1. A method for estimating external forces in a variable stiffness joint, characterized in that, Includes the following steps: Real-time acquisition of stiffness values; The obtained stiffness values ​​are substituted into the adaptive fusion coefficient model for calculation to obtain the adaptive fusion coefficients; the adaptive fusion coefficient model is as follows: in, Indicates joint stiffness The absolute value of the derivative, and All are positive numbers, satisfying ; External force estimation I is obtained based on the current information of the variable stiffness joint; External force estimation based on elastic deformation information and stiffness values ​​of variable stiffness joints II; Based on the adaptive fusion coefficient, external force estimation I, and external force estimation II, the external force estimate based on the fusion of current and elastic deformation information is obtained: in, For external force estimation I, For external force estimation II, For adaptive fusion coefficients.

2. The method for estimating external forces in a variable stiffness joint according to claim 1, characterized in that, The expression for the external force estimation II is: in, This indicates the stiffness value. For stiffness variables, This indicates elastic deformation inside the joint.

3. The method for estimating external forces in a variable stiffness joint according to claim 1, characterized in that, The method for obtaining external force estimation I based on the current information of the variable stiffness joint includes: Establish dynamic models for the position-driven module and the stiffness adjustment module: in, and These represent the moments of inertia of the position drive module and the stiffness adjustment module, respectively. and These represent the frictional forces of the position drive module and the stiffness adjustment module, respectively. and These represent the output torques of the position drive module and the stiffness adjustment module, respectively. and These represent the input torques of the position drive module and the stiffness adjustment module, respectively. and These represent the output angular displacements of the position drive module and the stiffness adjustment module, respectively. The moment of inertia in the dynamic model is obtained through dynamic parameter identification methods. , and friction , The accurate value; Measure the current information of the position motor and the stiffness adjustment motor. and The input torques of the position drive module and the stiffness adjustment module are calculated according to the following formula: in, and These represent the total torque constants of the position drive module and the stiffness adjustment module, respectively; The calculated input torques of the position drive module and stiffness adjustment module are substituted into the dynamic model of the position drive module and stiffness adjustment module for solution, so as to obtain the output torques of the position drive module and stiffness adjustment module. The sum of the output torques of the position drive module and the stiffness adjustment module is calculated as the external force estimate I: 。 4. The method for estimating external forces in a variable stiffness joint according to claim 1, characterized in that, The steps for obtaining the stiffness value in real time are as follows: Design a stiffness zone controller and calculate the expected value of stiffness variables based on the estimated external force and maximum external moment constraint at the previous moment. The expected value of the stiffness variable is applied to the motion control of the stiffness-adjusting motor to obtain the actual value of the stiffness variable; Substituting the actual values ​​of the stiffness variables into the stiffness model yields the updated stiffness values, where the stiffness model is a function of the stiffness variables.

5. The method for estimating external forces in a variable stiffness joint according to claim 4, characterized in that, When the variable stiffness joint is a lever-type variable stiffness joint, the designed stiffness region controller calculates the expected value of the stiffness variable based on the estimated external force and the maximum external moment constraint at the previous moment, including: (1) Design a stiffness region controller based on the Lyapunov barrier function; For lever-type variable stiffness joints, the Lyapunov barrier function used for stiffness region control is expressed as: in, Represents function variables, ; , , All are positive numbers and satisfy , ; This represents the estimated external torque; Represents a specific stiffness variable The maximum allowable external moment of the lower joint, and the stiffness variable related; The range of values ​​is ; (2) Based on the Lyapunov barrier function, the stiffness adjustment is divided into three regions: the stiffness increase region, the stiffness decrease region, and the constant stiffness region. (3) Based on the region involved by the Lyapunov barrier function, its derivative is expressed as: (4) Design a suitable To satisfy To ensure Located in the constant stiffness region, The expression is: in, Let be a constant representing the rate at which the stiffness of a lever-type variable stiffness joint transitions from a region of increasing stiffness or decreasing stiffness to a region of constant stiffness, and Greater than 0; (5) According to The definition will The derivative is expressed as: in, Represents the stiffness variable of a lever-type variable stiffness joint. The derivative; (6) After further sorting, the following results were obtained The expression: 。 6. The method for estimating external forces in a variable stiffness joint according to claim 5, characterized in that: When dividing stiffness adjustment into three regions based on the Lyapunov barrier function, the parameters of the Lyapunov barrier function are used as the basis. ; when At this time, it represents the region where stiffness is increased; when At this time, it is a region where stiffness decreases; when At that time, it is a region of constant stiffness.

7. A variable stiffness joint external force estimation device, characterized in that, include: The stiffness acquisition module is configured to acquire stiffness values ​​in real time. The external force estimation I acquisition module is configured to acquire external force estimation I based on the current information of the variable stiffness joint; The External Force Estimation II acquisition module is configured to acquire External Force Estimation II based on the elastic deformation information and stiffness values ​​of the variable stiffness joint; An adaptive fusion coefficient model obtains updated stiffness values ​​based on the actual values ​​of stiffness variables, wherein the adaptive fusion coefficient model is: in, Indicates joint stiffness The absolute value of the derivative, and All are positive numbers, satisfying ; The fusion module is used to obtain an external force estimate based on the fusion of current and elastic deformation information, using adaptive fusion coefficients, external force estimation I, and external force estimation II. in, For the estimation of external force I, For external force estimation II, For adaptive fusion coefficients.

8. The variable stiffness joint external force estimation device according to claim 7, characterized in that, The expression for the external force estimation II is: in, This indicates the stiffness value. For stiffness variables, This indicates elastic deformation inside the joint.

9. The variable stiffness joint external force estimation device according to claim 7, characterized in that, The stiffness acquisition module includes: The stiffness variable expectation value calculation module is configured as a design stiffness zone controller, which calculates the expectation value of stiffness variables based on the estimated external force value and the maximum external moment constraint at the previous moment. The stiffness variable actual value acquisition module is configured to apply the expected value of the stiffness variable to the motion control of the stiffness adjustment motor to obtain the actual value of the stiffness variable; The stiffness model is configured to obtain updated stiffness values ​​based on the actual values ​​of stiffness variables, where the stiffness model is a function of the stiffness variables.

10. The variable stiffness joint external force estimation device according to claim 7, characterized in that, The external force estimation I acquisition module includes: The dynamics model building module is configured to create dynamics models for the position-driven module and the stiffness adjustment module: in, and These represent the moments of inertia of the position drive module and the stiffness adjustment module, respectively. and These represent the frictional forces of the position drive module and the stiffness adjustment module, respectively. and These represent the output torques of the position drive module and the stiffness adjustment module, respectively. and These represent the input torques of the position drive module and the stiffness adjustment module, respectively. and These represent the output angular displacements of the position drive module and the stiffness adjustment module, respectively. The parameter identification module is configured to obtain the moment of inertia in the dynamic model through a dynamic parameter identification method. , and friction , The accurate value; The external force calculation module is configured to calculate the current information of the measured position motor and stiffness adjustment motor. and The input torques of the position drive module and the stiffness adjustment module are calculated using the following formula: in, and These represent the total torque constants of the position drive module and the stiffness adjustment module, respectively; The output torque calculation module is configured to substitute the calculated input torques of the position drive module and stiffness adjustment module into the dynamic models of the position drive module and stiffness adjustment module for solution, thereby obtaining the output torques of the position drive module and stiffness adjustment module; and to calculate the sum of the output torques of the position drive module and stiffness adjustment module as the external force estimate I. 。

Images