A method for dynamic characteristic identification of robot multi-dimensional force/torque sensor

Abstract

The invention discloses a method for identifying the dynamic characteristics of a robot multi-dimensional force / torque sensor based on time-domain iterative averaging. Moment sensor excitation force and displacement response signal; 2. Mathematically model the impulse response function of the multidimensional force / torque sensor; 3. Calculate the average error value W of the displacement response sequence after M times of excitation of the multidimensional force / torque sensor; 4. Calculate the recursive formula eh(t) of the impulse response function error through the average error; 5. Calculate the impulse response function according to the impulse response function error; 6. Identify the dynamic characteristics of the calculated impulse response function to obtain the multidimensional force / The natural frequencies, damping ratios, and mode shapes of multidimensional force / torque sensors for torque sensors. The invention can directly extract the impulse response function of the multi-dimensional force / torque sensor in the time domain, thereby improving the accuracy of identifying the dynamic characteristics of the multi-dimensional force / torque sensor.

Description

technical field [0001] The invention relates to an identification technology, in particular to a fast identification technology of time-domain dynamic characteristics of a robot multi-dimensional force / torque sensor. Background technique [0002] As a very important intelligent robot sensor, the multi-dimensional force / torque sensor is installed between the arm and the gripper of the manipulator to obtain high-precision force and torque information in real time, so the dynamic characteristics of the multi-dimensional force / torque sensor directly affect the force and torque information. Accuracy obtained. The dynamic characteristics of multidimensional force / torque sensors are usually described by dynamic mathematical models, and the mathematical models of multidimensional force / torque sensors are established to obtain their dynamic characteristics. The commonly used modeling method is the system identification method using the observation data of the stimulus response. The ...

AI Technical Summary

Problems solved by technology

The commonly used modeling method is the system identification method using the observation data of the stimulus response. The most commonly used method is to obtain the frequency response function by Fourier transform in the frequency domain, and obtain the impulse response function by Fourier inverse transform. However, the Fourier The leaf transform has the characteristics of frequency aliasing and energy leakage, which has a very serious impact on the extraction of the dynamic characteristics of the multi-dimensional force / torque sensor; the dynamic characteristics of the multi-dimensional force / torque sensor are studied in the time-frequency domain by wavelet transform, although the accuracy has been improved to a certain extent. , but the choice of wavelet base is a crucial factor affecting the extraction of the dynamic characteristics of multidimensional force / torque sensors. In addition, wavelet transform is essentially the same as Fourier transform, and the shortcomings of Fourier transform also exist in wavelet transform; multidimensional force / torque in the time domain The identification of the dynamic characteristics of the torque sensor usually calculates the inverse matrix of the excitation signal based on the Duhamel integral. This method has a large amount of calculation, low efficiency, and is easily disturbed by noise. The accuracy of identifying the dynamic characteristics of the multi-dimensional force / torque sensor is low.

Images