Quantitative Methods in Defense and National Security 2007

Multi-threat Containment with Cooperative Autonomous Agents
Shanchieh Jay Yang, (Department of Computer Engineering, Rochester Institute of Technology), jay.yang@rit.edu , and
Bhushan Mehendale, (Rochester Institute of Technology), bhushan@bhushan.in


The study of sensors and robotics has move beyond optimizing individual system performance. A key focus today is on overcoming challenges of utilizing a large number of cost-effective, autonomous, cooperative agents (sensors or robots). This work will lay out a few key open problems in the field, with a focus on the multi-threat containment problem. The multi-threat containment problem asks a set of autonomous agents to engulf and observe occurring threats aided by only local sensing capabilities. Previous related work has dealt with robot formations and single threat containment, in which cases a single target exist throughout the containment process. In order to accommodate multiple threats occurring at random times and expires, autonomous robots need to dynamically adjust their target threats to ensure containments. A potential-field based approach is taken in this work to enable distributed and multiple threat containment and collision avoidance. Each agent in the field will independently form its own view (the potential field) of the surrounding; this view will be periodically updated and used to determine the agent's direction and velocity in traveling to engulf the target threats. A combination of quadratic potential functions is used to model the potential field seen by the agents, which choice simplifies the computational complexity and the parameter design. The simplicity of the proposed algorithm, Multiple Threat Containment Algorithm (MUTCA), is expected to enable its realization on cost-effective robots (~$200). MUTCA has been simulated with different threat occurrence settings as well as robot capabilities. The talk will demonstrate the simulation results, showing the benefits and the limitations of MUTCA, and conclude with a discussion on future directions of this research.

Take me back to the main conference page.