ADALINE Stabilization of an Inverted Robot Arm

By
Peter C. Jones
and
Scott Tepavich

Advised by
Dr. T. L. Stewart & Dr. G. Dempsey


Abstract

A useful control tool is one that can adapt to variances in a complex system. One system that would benefit from such adaptive control is an inverted robot arm, the exact nature of which is not known or can be changed (length, mass, mass distribution, disturbances). This project is the implementation of an ADALINE (Adaptive Linear Neuron) neural network learning control system on a Motorola DSP56002 Digital Signal Processor to stabilize the robot arm to a vertical position. The control network is trained with the adjustment signals used to manually stabilize the robot arm and the resulting robot arm angles. Once trained, the network will stabilize the robot arm without manual input.


Papers Available


Copyright © 1997 by Peter C. Jones and Scott D. Tepavich.
All Rights Reserved.