For the accuracy requirement of the motion planning and control in the industrial application of parallel robot, proposes an particle swarm optimization algorithm based on adaptive diversity feedback (APSO) to improve the accuracy of parallel robots. Use the closed loop vector to calculate the robot’s kinematics solution, and the application of APSO turn the structure parameters optimization problem into an optimization problem of nonlinear system, by calibrate the robot drive motor rotation, the machine get an ideal accuracy in the working space. Test on SRBD Delta parallel robot, and the simulation results can reach the level of 10-3mm, the validity of the method is verified.