A discrete-time learning control for robotic manipulators is studied system theoretically with a pulse transfer functions.
Firstly, the learning stability condition for the discrete-time learning control is derived in frequency domain. From that condition, we derive the new learning stability condition which is applicable for the case of which output number is larger than input number. This is more common as in the case of voltage input position and velocity outputs.
Secondly, we examine the learning algorithm with the position signal. When the sampling period is small, the learning stability is degraded because of the system's zeros which are generated by discretizing
Thirdly, we examine the learning algorithm with position and velocity signals. In this case, we can stabilize the learning control systems which are unstable in the case of the learning with only the position signal.
Finally, we show the simulation results on the trajectory control of robotic manipulators using the discrete-time learning control. The simulation results agreed well with the analytical ones.
抄録全体を表示