Teaching neural networks to guide a vehicle through an obstacle course by emulating a human teacher

An interesting application of neural networks in man-machine systems is the identification of human operator behavior by observation. A network trained to emulate human strategies may subsequently be used to monitor the human online, to detect possibly dangerous system states and set alarms, or to take over automatically in cases of emergency. The feasibility of this approach is studied using the navigation of a vehicle through an obstacle course as a paradigm. Network architectures for vehicle control and path optimization are developed and tested. This includes a novel technique called functional-link net. The functionality of a demonstrator prototype is described