Object surveillance using reinforcement learning based sensor dispatching

摘要: This paper outlines an approach to the coordination of multiple mobile sensors for surveillance a single moving target. A real-time dispatching algorithm is used select and position groups in response observed object motion. The aim provide robust, high-quality data while ensuring that system can react unexpected manoeuvres. Sensors are assigned collect at specific points on trajectory. strategy learned via reinforcement learning control sensor poses with respect these points. In using strategy, each adopts egocentric view state determine most appropriate action. Simulations demonstrate performance RL-based dispatcher, comparison similar static-sensor systems.