Research on Dynamics Modeling and Obstacle Avoidance Optimization of Hyper-Redundant Robotic Arm Based on Deep Learning

A dynamic model identification method for semi-parametric hyper redundant manipulator based on iterative optimization and neural network compensation is proposed. Firstly, the motion modeling and optimal control parameter set of the hyper-redundant robot system are studied in this paper, and the nonlinear friction mathematical model of the robot motion pair is constructed, and the iterative optimization algorithm is adopted to realize the accurate recognition of the robot motion inertia and the motion characteristics of the moving pairs. The BP neural network based on training samples is trained, and the semi-parametric dynamic modeling method of over-redundancy robot is obtained. Simulations demonstrate that compared to traditional methods, the proposed deep learning approach significantly improves the alignment of the robotic arm's joint angles with preset expectations, maintaining a safe distance of 0.2 to 0.3 meters from obstacles. These results underscore the method's superiority in enhancing both the accuracy and safety of robotic arm operations in complex environments. The verification outcomes confirm the method's efficacy and supremacy in identification tasks.