ADAPTIVE NEURAL-NET CONTROLLER SYSTEM FOR UNDERWATER VEHICLE

摘要

This paper describes the concept and example of an adaptive neural-net controller system for underwater vehicle. The system has two independent parts in which necessary processes for data sampling, robot control, identification of dynamics and controller adaptation are executed parallel with each other process. The data for each processes such as synaptic weights of controller are communicated between related processes asynchronously. The system is implemented onto a robot's computer system with parallel processing capability, and adaptability of the controller system is investigated by tank tests.