Abstract:The forward kinematics model of the 6R serial robot was set up by using conventional D-H modeling method. when model has errors,the inverse solution of trajectory planning target contains errors, so the actual running trajectory can not satisfy the 〖JP2〗precision of robot operation. The error between the target pose and actual pose was proposed as iterative target. Based on the Levenberg-〖JP〗Marquardt method, the iterative solution precision was obtained through iterative optimization.with the model parameters containing errors, and the corrected inverse solution of joint angle was output , which can make the actual motion of the robot close to the target pose of trajectory planning with the desired operating precision.Simulation results show that the algorithm can solve optimization of inverse solution for complex trajectory planning, and avoid the high precision calibration and parameter identification of the kinematic model, which has the practically application value.