Forward and reverse trajectory features fused GRU neural network for hysteresis characteristics modeling of joint

To address the problem of high-precision modeling of the complex hysteresis characteristics exhibited by the flexible joints of low-cost and lightweight industrial robots without load torque sensors, the motor drive current is used to indirectly reflect the change of load torque, and the relationship between drive current and torsion angle is employed to describe the hysteresis characteristics of flexible joints, the forward and reverse trajectory features -based GRU neural network hysteresis model is proposed. The features specific to the forward and reverse trajectory features in the joint hysteresis characteristics are fused into the GRU neural network. The Kalman filter-based current increments are used to extract the features of the forward and reverse trajectory to describe the multi-valued characteristics exhibited by the forward and reverse trajectory features in the current-torsion angle hysteresis characteristics. The historical prediction results are taken as the new information input model to construct a dynamic GRU neural network hysteresis model with memory capability and nonlinear mapping ability. The experimental results verify that the proposed flexible joint hysteresis model has good prediction ability and high model accuracy.