Motion control for vehicle with unknown operating properties -on-line data acquisition and motion planning

This paper describes a sequence of methods for controlling the vehicle with unknown operating properties, such as an air cushion vehicle, a submarine and an airship. We took up Air Cushion Vehicle (ACV) as an example of such a vehicle, and studied its motion planning based on acquired data. ACV in this paper has three simple unit motions (left-turn, right-turn and an advance), and can execute one of the unit motions optionally. After the execution, ACV measures its own position by itself and acquires data of how it moved in the real environment. Then it plans the motion that arrives at a target position using the acquired data. In this paper, we propose a sequence of these approaches and show the effectiveness of them by experiments.