Search

CN-118636127-B - Joint torque estimation method of joint type robot based on discrete non-smooth observer

CN118636127BCN 118636127 BCN118636127 BCN 118636127BCN-118636127-B

Abstract

The invention discloses a joint type robot joint torque estimation method based on a discrete non-smooth observer, which expands a term containing the robot joint torque into 3 variables of a system according to a joint type robot dynamics model, designs a discrete non-smooth observer algorithm based on a discrete non-smooth control theory, and estimates the robot joint torque and angular velocity values. The discrete non-smooth observer algorithm of the invention avoids the noise influence of the traditional differential method when estimating the torque and the angular velocity of the robot joint, and the observer of the invention only takes the position of the robot joint as an input signal when designing, thereby reducing the information requirement on an actual system.

Inventors

  • DU HAIBO
  • CONG YONGZHENG
  • FANG LIANDI
  • LUO YUEYUE
  • ZHU WENWU
  • ZHU MIN

Assignees

  • 合肥工业大学智能制造技术研究院

Dates

Publication Date
20260512
Application Date
20240524

Claims (7)

  1. 1. A joint type robot joint torque estimation method based on a discrete non-smooth observer comprises the steps of firstly reconstructing a robot dynamics model, expanding a term containing the robot joint torque into 3 system variables theta 1 、θ 2 and theta 3 ,θ 1 =theta, Theta 3 ＝M -1 (θ 1 )τ;θ∈R n×1 ,τ∈R n×1 , n is the number of the robot joints, theta is the position vector of the robot joints, Is the derivative of theta, tau is the torque of the robot joint, M (theta) epsilon R n×n , M (theta) is the inertial matrix of the robot, and M -1 (theta) is the inverse matrix of M (theta); Setting a torque observer, and then calculating an estimated value of the angular velocity of the robot joint at the moment of the specification based on the reconstructed robot dynamics model and the torque observer And joint torque estimation The torque observer is designed as a discrete non-smooth torque observer, and the formula is as follows: wherein k 1 ,k 2 ,k 3 ,k 4 ,k 5 ,k 6 each represent a positive gain, For an estimate of θ 1 ,θ 2 ,θ 3 , T is the sampling period of the discrete non-smooth observer, T k represents the kth time, T k+1 represents the k+1th time, θ 1 (t k ) represents the sample value of θ 1 at the kth time, Respectively represent the estimated value of theta 1 ,θ 2 ,θ 3 at the kth time, Respectively representing estimated values of theta 1 ,θ 2 ,θ 3 at the (k+1) th moment, sig and sgn being functions, and f c ∈R n×n being a coulomb friction coefficient of the robot; Is that Is used for the inverse matrix of (a), Taking value for robot joint position vector The robot inertia matrix; is a calculation item; Wherein M -1 (θ 1 ) represents an inverse matrix of M (theta 1 ), M (theta 1 ) is a robot inertia matrix when the robot joint position vector takes a value theta 1 , C (theta 1 ,θ 2 )∈R n×n is a robot centripetal force and coriolis force vector term, G (theta 1 )∈R n×1 is a robot gravity term vector; f v ∈R n×n is a robot viscosity friction coefficient, and f b ∈R n×1 is a robot friction offset value).
  2. 2. The method for estimating joint torque of an articulated robot based on a discrete non-smooth observer according to claim 1, wherein the reconstructed robot dynamics model is: Γ(θ 1 ,θ 2 )＝M -1 (θ 1 )(C(θ 1 ,θ 2 )θ 2 +G(θ 1 )+f v ·θ 2 +f b ) Wherein, the As a derivative of theta 1 , Is the derivative of theta 2 , gamma (theta 1 ,θ 2 ) is a calculation term, M -1 (θ 1 is an inverse matrix of M (theta 1 ), C (theta 1 ,θ 2 )∈R n×n is a centripetal force and coriolis force vector term of the robot, G (theta 1 )∈R n×1 is a gravity term vector of the robot, f v ∈R n×n is a viscous friction coefficient of the robot, and f b ∈R n×1 is a friction offset value of the robot.
  3. 3. The method for estimating joint torque of articulated robot based on discrete non-smooth observer according to claim 1, wherein the estimated value of the angular velocity of the joint of the robot at a given moment in time The calculation formula of (2) is as follows: Represents the estimated value of the angular velocity of the robot joint at the kth moment The estimated value of θ 2 at the kth time is shown.
  4. 4. The method for estimating joint torque of an articulated robot based on a discrete non-smooth observer according to claim 1, wherein the estimated value of the joint torque of the robot at a given moment in time The calculation formula of (2) is as follows: estimated value representing robot joint torque τ at kth time Indicating robot joint position vector value The inertial matrix of the robot at the time of the process, Represents an estimate of theta 1 at the kth time, The estimated value of θ 3 at the kth time is shown.
  5. 5. A system for implementing the discrete non-smooth observer based articulated robot joint torque estimation method according to any one of claims 1-4, wherein the system comprises a robot dynamics model, a torque observer, a sampling module, and a processor; The sampling module is used for acquiring a robot joint position vector theta, the processor is connected with the sampling module, the processor substitutes the theta acquired by the sampling module into a robot dynamics model and solves the model in combination with the torque observer so as to calculate an estimated value of the angular velocity of the robot joint at a specified moment And joint torque estimation
  6. 6. An articulated robot joint torque estimation system based on a discrete non-smooth observer, comprising a memory, wherein the memory stores a computer program which, when executed, is configured to implement the articulated robot joint torque estimation method based on a discrete non-smooth observer as defined in any one of claims 1-4.
  7. 7. The discrete non-smooth observer based articulated robot joint torque estimation system according to claim 6, further comprising a processor coupled to the memory, the processor for executing the computer program to implement the discrete non-smooth observer based articulated robot joint torque estimation method according to any one of claims 1-4.

Description

Joint torque estimation method of joint type robot based on discrete non-smooth observer Technical Field The invention relates to the technical field of robot power estimation, in particular to a joint torque estimation method of a joint type robot based on a discrete non-smooth observer. Background With the development of industrial automation technology, industrial robots are widely used in various industrial fields. However, in some special applications, contact with the surrounding environment and the operation object is required, and at this time, the interaction force between the robot and the external environment needs to be considered, where the conventional position control method is not applicable, and often a force control manner is required. The moment value of each joint of the robot is key information in force control, and the moment of the joint of the robot can be obtained most directly and accurately through the force sensor, however, the force sensor is expensive, so that the development and design of the moment observer with simple structure and excellent performance have important significance. Many scholars estimate using state observers, such as adaptive observers, extended state observers, non-smooth observers, etc. The non-smooth observer has high estimation speed and high precision due to good robustness, and is widely focused, and a plurality of students estimate the joint moment of the industrial robot by adopting the non-smooth observer and obtain good effects, but the input signal of the observer comprises a robot joint angle signal and an angular velocity signal, and the angular velocity of the robot joint is obtained through joint position differential filtering, so that the calculated amount of an industrial robot control system is increased. At present, a moment observer using only a robot joint position as an input signal has little research, and the known moment observer using only a robot joint position as an input signal has low estimation accuracy and low calculation speed. Disclosure of Invention In order to overcome the defect that a moment observer in the prior art is difficult to meet the requirement of the motion precision of the robot, the invention provides a joint type robot joint torque estimation method based on a discrete non-smooth observer, which solves the problem of estimating the robot joint torque value under the condition of no external torque sensor and improves the motion control precision and stability of the robot. The invention provides a joint type robot joint torque estimation method based on a discrete non-smooth observer, which comprises the steps of firstly reconstructing a robot dynamics model, expanding a term containing the robot joint torque into 3 system variables theta 1、θ2 and theta 3,θ1 =theta,Theta 3＝M-1(θ1)τ;θ∈Rn×1,τ∈Rn×1, n is the number of the robot joints, theta is the position vector of the robot joints,Is the derivative of theta, tau is the torque of the robot joint, M (theta) epsilon R n×n, M (theta) is the inertial matrix of the robot, and M -1 (theta) is the inverse matrix of M (theta); Setting a torque observer, and then calculating an estimated value of the angular velocity of the robot joint at the moment of the specification based on the reconstructed robot dynamics model and the torque observer And joint torque estimation The torque observer is designed as a discrete non-smooth torque observer, and the formula is as follows: wherein k 1,k2,k3,k4,k5,k6 each represent a positive gain, For an estimate of θ 1,θ2,θ3, T is the sampling period of the discrete non-smooth observer, T k represents the kth time, T k+1 represents the k+1th time, θ 1(tk) represents the sample value of θ 1 at the kth time,Respectively represent the estimated value of theta 1,θ2,θ3 at the kth time,Respectively representing estimated values of theta 1,θ2,θ3 at the (k+1) th moment, sig and sgn being functions, and f c∈Rn×n being a coulomb friction coefficient of the robot; Is that Is used for the inverse matrix of (a),Taking value for robot joint position vectorThe robot inertia matrix; is a calculation item; Wherein M -1(θ1) represents an inverse matrix of M (theta 1), M (theta 1) is a robot inertia matrix when the robot joint position vector takes a value theta 1, C (theta 1,θ2)∈Rn×n is a robot centripetal force and coriolis force vector term, G (theta 1)∈Rn×1 is a robot gravity term vector; f v∈Rn×n is a robot viscosity friction coefficient, and f b∈Rn×1 is a robot friction offset value). Preferably, the reconstructed robot dynamics model is: Γ(θ1,θ2)＝M-1(θ1)(C(θ1,θ2)θ2+G(θ1)+fv·θ2+fb) Wherein, the As a derivative of theta 1,Is the derivative of theta 2, gamma (theta 1,θ2) is a calculation term, M -1(θ1 is an inverse matrix of M (theta 1), C (theta 1,θ2)∈Rn×n is a centripetal force and coriolis force vector term of the robot, G (theta 1)∈Rn×1 is a gravity term vector of the robot, f v∈Rn×n is a viscous friction coefficient of the robot, and f b∈Rn×1