Partitioning Strategies and Task Allocation for Target-tracking with Multiple Guards in Polygonal Environments

Hamid Emadi,Tianshuang Gao,Sourabh Bhattacharya

DOI: https://doi.org/10.48550/arXiv.1611.05092

2016-11-16

Abstract:This paper presents an algorithm to deploy a team of {\it free} guards equipped with omni-directional cameras for tracking a bounded speed intruder inside a simply-connected polygonal environment. The proposed algorithm partitions the environment into smaller polygons, and assigns a guard to each partition so that the intruder is visible to at least one guard at all times. Based on the concept of {\it dynamic zones} introduced in this paper, we propose event-triggered strategies for the guards to track the intruder. We show that the number of guards deployed by the algorithm for tracking is strictly less than $\lfloor {\frac{n}{3}} \rfloor$ which is sufficient and sometimes necessary for coverage. We derive an upper bound on the speed of the mobile guard required for successful tracking which depends on the intruder's speed, the road map of the mobile guards, and geometry of the environment. Finally, we extend the aforementioned analysis to orthogonal polygons, and show that the upper bound on the number of guards deployed for tracking is strictly less than $\lfloor {\frac{n}{4}} \rfloor$ which is sufficient and sometimes necessary for the coverage problem.

Robotics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to use multiple mobile robots (called guards) carrying omnidirectional cameras in a polygonal environment to track an intruder with a limited speed. Specifically, the paper proposes an algorithm that divides the environment into smaller polygonal regions and assigns a guard to each region to ensure that the intruder is seen by at least one guard at all times. In addition, the paper also introduces the concept of "dynamic regions" and proposes an event - triggering strategy for guards to track the intruder based on this. The main contributions of the paper include:

1. Proposing a deployment algorithm that requires a strictly smaller number of guards than $\left\lfloor \frac{n}{3} \right\rfloor$, which is a sufficient and sometimes necessary condition for covering a simple polygon.
2. Introducing the concept of "dynamic regions" and proposing an event - triggering strategy for guards to track the intruder based on this.
3. Giving an upper bound on the maximum speed required for guards to successfully track the intruder, which depends on the speed of the intruder, the path graph of the guards, and the geometry of the environment.
4. Extending the above analysis to orthogonal polygons and proving that in this case, the number of deployed guards is strictly less than $\left\lfloor \frac{n}{4} \right\rfloor$, which is also a sufficient and sometimes necessary condition for the covering problem.

Through these contributions, the paper aims to improve the efficiency and reliability of target tracking in polygonal environments.