RI: Small: Robust motion planning for multirobot target tracking

点击翻译按钮获取中文摘要

Autonomous mobile robots have recently been used for tasks such as surveillance and intruder detection. Coordinating a network of such robots, though, is a difficult task, especially if the people are trying to avoid detection. This research aims to develop a network of autonomous aerial vehicles that can detect and track intruders in farmlands. In the face of recent incidents related to agricultural espionage and biotech piracy in open farms, a robotic intrusion detection system would be helpful to farmers. The techniques developed would also be applicable in areas such as air traffic control, surveillance of transportation networks, and patrol of coastal areas. Through educational and outreach activities, the project aims to create a highly skilled workforce in the growing areas robotics and cyber-security. The proposed research tackles the problem of designing robust motion plans for multirobot systems deployed for tracking unpredictable and adversarial targets. A serious challenge in such problems is the lack of information about the targets' future actions. Based on tools from differential game theory, this research develops target allocation and trajectory generation algorithms for a team of observer robots to track multiple intruders, with guarantees on tracking performance. Using a bottom-up approach, a motion planner is proposed for multirobot target tracking that can handle sensing and motion constraints posed by obstacles in the environment, variable number of targets, and noisy sensor and actuator measurements. The proposed research includes testing and validation of the tracking planner on a multi-quadrotor testbed in the PI's lab to detect unauthorized intrusion in agricultural farmlands.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.