Research On The Nonlinear Control Of High-speed Delta Robot

By its easy and elegant structure as well as fast and accurate motion, the Delta robots get a huge business success. However, as the increasing demand of robot speed, the regular PID controller shows its limitations, which are higher tracking error, more fluctuation of torque and the resulting bigger motor vibration and noise. The main causes of this phenomenon, driven by detailed research, are believed to be: when the robot moves in high speed, its inertial forces become enormous and the couplings between joints become severe, while the PID controller that is not based on model, is hard to tune a satisfied parameter for the highly coupled system. Therefore, this paper considers to introduce the system dynamics and use the nonlinear controllers to compensate the high nonlinear character raised in the high speed application of Delta robot.In this paper, based on the forward, inverse kinematics obtained by geometric method,and the Jacobian and Hessian matrix, a more perfect dynamic model of Delta robot was established by using the generalized virtual work principle. This model takes the Coulomb and viscous friction of the drive joints into consideration. This model also adds one item to compensate the inertial error introduced by the simplification of real links.Then based on the modified Delta robot dynamic model, a Computed Torque Controller(CTC) and a Torque Feedforward Controller(TFC) that used the desire states to substitute the real states were derived in this paper. Considering the adverse impact of the payload effect under high speed circumstance, a Self Adaptive Controller(SAC) and a Desired Compensation Adaptive Law(DCAL) controller were designed, then the stablity proof of the SAC was presented.Finally, after measuring and identifying the dynamic parameters used in control algorithms, A x PC-based Delta robot control system was built, to explore the performances of the controllers derived in this paper. The experiments show that, because of the measurement noise in the joint angular speed, the CTC have a very limited improvement to the PID controller. The TFC, however, because of using the desired states instead of the measured, having a better performance, brings the torque impact. To solve this problem,the first order inertial element was added after the TFC, which effectively smoothed the peaks in the torque. By using this modified algorithm, the time to execute the standard test path Adept Motion can be shortened to 204 ms, along with the improvement of tracking accuracy and torque fluctuation. This paper also experimented the SAC, which shows that this algorithm can identify the uncertain parameters successfully. However, the experiments also revealed the poor robustness of the SAC, which relies heavily on the tracking accuracy. Therefore, under the present circumstance, it is difficult to deploy in the high speed application of Delta robot.