Knowledge acquisition for collision avoidance using fuzzy neural networks

Acquisition of control rules have been actively studied for improving the performances of robot control systems. It has been difficult to obtain the collision avoidance rules which coincide with the operator´s knowledge. This paper shows that the operators´ avoiding rules can be acquired directly from data, which the operators probably observe, using a fuzzy neural network (FNN). This paper also shows that the FNN can obtain portions of the fuzzy rules for the inferences of the static and dynamic degrees of danger and the decision table to avoid the moving obstacle.