Simulation of Fully Autonomous Control of Unmanned Air Vehicles for Maritime Surveillance

摘要

As the usage of unmanned systems become more prevalent, defence departments around the world are looking for new modes of human-system interaction that increase the level of autonomy of the vehicle, while maintaining operator control and trust. Defence R&D Canada is currently engaged in a research program on the simulation of Unmanned Aerial Vehicles (UAVs) and methods to increase autonomous capabilities. In this paper we describe a sensor-based reactive control system for a UAV which can perform a mission autonomously. The primary sensor is machine vision, however, the focus on this paper is that of implementing reactive control for UAV autonomy. The mission, safety and contingency planning are accomplished through a set of prioritized parallel controllers to produce a composite emergent control action. The parallel controllers are implemented using finite state machines and state diagrams. The development and evaluation of this autonomous controller is conducted on an engineering simulator being constructed at Defence R&D Canada in Ottawa. The design and evaluation mission is an end-phase interception mission for maritime surveillance. Simulation trials are presented demonstrating the ability of UAV to adapt to varying situations.