Research On Gravity Compensation Of Robot Arm Based On Model Learning

In this paper, an algorithm based on model learning to calibrate gravity compensation parameters is proposed. This method can be used to solve the gravity term of the mechanical arm by linear regression method based only on the structure of the arm and the sampling data. The position of the gravity center and the geometric center of each mechanical arm link is not needed. It solves the problem that the model derived from the traditional dynamic modeling method produces a large deviation from the actual model. Theoretical analysis shows that the joint angle and joint driving torque can accurately regress the relationship between the barycentric coordinates of the link and the mass of the connecting rod in the Cartesian coordinate system, thus realizing the gravity compensation of the manipulator. The optimal solution on the optimization trade-off curve is presented and discussed.