A three-neuron controller. II

For pt. I see IEEE Conference on Systems, Man and Cybernetics (1995). In that part, the biological plausibility of neural control circuits which consist of two input and one output neurons (a three-neuron controller or TNC) was discussed. The task of backing up a truck to a loading dock was selected as a benchmark and the results obtained in this task by such a controller compared to the results obtained by a fuzzy controller. Since this comparison was favorable to the TNC, we now continue exploring the capabilities of such control. In this second paper we explore the use of equations that are more biologically consistent, and demonstrate their use in the truck (cab) and trailer backer-upper. These equations can be obtained by an extension to McClelland and Rumelhart's interactive activation and competition equations (1988) or by simplifying the equations in Alexander's earlier work (1991). The new equations will be called ENIAC (extended interactive activation and competition). The approach to using TNC's in controlling a truck and trailer backing up is based on the method used by Kosko (1992), but we have greatly simplified the control of the cab or tractor. Control is seen to be on a par with a fuzzy controller.