Collaborative Risk-Avoidance Strategy for Multirobot Information Gathering in Adversarial Environment

Multirobot information gathering in adversarial environments, where enemies can destroy detected robots, presents unique challenges not addressed by existing algorithms designed for safe settings. The advent of multirobot systems enables collaborative risk-avoidance behavior, exemplified in pursuit-evasion scenarios, where concentrated robot groups enhance their ability to evade and overcome pursuing enemies. However, current multirobot information gathering algorithms have not yet integrated this collaborative risk-avoidance model, potentially leading to robot damage and reduced efficiency. This study adopts a decomposition-and-grafting mechanism to separate the problem into two weakly coupled subproblems: task execution and task allocation. For task execution, we propose an exact algorithm based on the branch-and-pricing method. Our task execution algorithms are seamlessly integrated with a novel task map composition algorithm designed to identify high-utility solutions with minimal computational overhead, addressing the task allocation problem. Extensive simulations demonstrate that our algorithms significantly outperform benchmarks by increasing the number of collaborative risk-avoidance activities conducted by robots during information gathering tasks in adversarial environments. This research advances multirobot information gathering by incorporating collaborative risk-avoidance, enhancing robot survivability and efficiency in hazardous settings.