Experimental evaluation of real-time model-based control of a 3-DOF closed-chain direct-drive mechanism

The authors present results on the performance of a 3 d.o.f. closed-chain direct-drive mechanism controlled by a high-performance real-time computer using a variety of model-based control techniques. Experimental data are used to compare the tracking performance of the computed torque control method, the resolved acceleration control method, the inverse Jacobian control method, and the individual joint PID control method over a variety of trajectories. Implementation in real-time was possible since the 3D processor robot control system was able to perform all of the required kinematic, dynamic, and control computations at a sampling rate of under 2 ms. The results presented show that the model-based methods achieved two to four times better performance than the PID control method