Adaptive robot control using a processor network
F. Azhar, David A. Fraser · 2002
Control of a revolute robot manipulator has been distributed over a transputer network. Feedforward and feedback position control filters for the three major axes have been implemented as neural nets. With one hidden layer and trained using back-propagation. The feedforward filter continues to adapt during operation and shows a steady improvement in control quality, and the ability to adapt to changes in the system dynamics.>