Toward Real-Time Autonomous Target Area Protection: Theory and Implementation

This brief considers the target guarding problem (TGP) with a single pursuer F, a single evader E, and a stationary target T. The goal of F is to prevent E from capturing T, by intercepting E as far away from T as possible. An optimal solution to this problem, referred to as a command to optimal interception point (COIP), was proposed recently. This guidance law requires the positions of the agents involved. Typically, aerial sensors, such as GPS, used for obtaining these data may not always perform robustly on the field, thereby reducing the autonomy of the vehicles. The computational complexity of the expressions in the COIP law also makes it difficult for a real-time implementation. Here, the TGP is revisited and the optimal solution is reformulated to expressions that are suitable for autonomous systems with ranging sensors mounted on them. These expressions also allow for seamless real-time implementation in robotic hardware. The reformulation enables the optimal solution to be coded as a lookup table requiring minimal memory to further increase the speed of computations. An experimental setup with mobile robots is then used to validate the claims. The case of T lying in E's dominance region is considered a lost game for F. However, this is true only if E plays optimally. If E plays suboptimally F stands a chance to win the game. This case, which has not been analyzed earlier, is also discussed in this brief, and an optimal strategy for F is presented.