Team-aware Robotic Demining Agents for Military Simulation

Although real-world solutions for robotic demining have not as yet been demonstrated, multirobot systems will clearly yield significant benefits over a single robot, given the extremely hazardous nature of the domain [JP1999]. Having multiple robots adds a desirable redundancy to the system, as well as potentially reducing the time required to clear an area if the robots are deployed effectively. In this paper, we describe a software implementation of a multirobot demining system that was deployed as part of a larger multiagent system, AgentStorm, to assist human commanders in command and control scenarios. Simulated robotic demining agents coordinate to breach a path through a minefield; demining is modeled as a distributed optimization problem in which the robots strive to minimize an abstract cost function. Although researchers have drawn comparisons between robot demining and foraging [Goldberg & Mataric1997], path breaching cannot be modeled as a simple foraging task because the goal of the exercise is not to optimize on number of mines removed but to assure coverage of a connected path.