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.