A robot motion error parameter optimization method

Abstract

The application discloses a robot motion error parameter optimization method, and belongs to the field of robots. The method comprises the following steps: controlling a robot to perform a plurality of preset motion postures; collecting actual end position data under each posture by means of an end measurement device; combining theoretical position data output by a forward kinematics model to obtain end position error data under each posture; realizing collaborative accurate quantification of error parameter contribution degree evaluation and coupling relationship identification through neural network global sensitivity analysis and double coupling measurement, and overcoming the defect that traditional sensitivity analysis can only evaluate independent influence; based on a double measurement mechanism of linear coupling similarity and high-order interactive coupling strength, complex coupling characteristics among parameters under different motion postures are fully captured, complete and reliable data basis is provided for subsequent parameter grouping and staged optimization, and identification deviation caused by neglecting parameter coupling is avoided from the root, and the reduction accuracy of parameter optimization results to actual robot structure errors is improved.