Abstract:
A multi-objective optimization algorithm based on manifolds is proposed to solve the shortcomings of large operand and poor real-time performance of the interpolation optimization algorithm for the inverse kinematics solutions of redundant robots. The solution space of inverse kinematics of the redundant robot is regarded as a smooth manifold, the dimensionality reduction analysis is conducted in the position workspace manifolds and posture workspace manifolds. The optimal inverse solutions of the redundant robot are obtained according to the proposed optimization objective function. The optimal performance indexes may be contradictory in multi-objective optimization of the redundant robot, and thus the weights should be added to each index according to their priorities to realize global optimization of the redundant robot. Generally, the optimal inverse solution is not a single solution, but an optimal solution manifold. Finally, the rationality of the proposed method is demonstrated by inverse kinematics simulation with the plane S-shaped inlet.