Coverage control in unknown environments using neural networks
This paper presents coordination algorithms for groups of mobile agents performing deployment and coverage tasks. As an important modeling constraint, we assume that each mobile agent has a limited sensing or communication radius. Based on the geometry of Voronoi partitions and proximity graphs, we analyze a class of aggregate objective functions and propose coverage algorithms in continuous and discrete time. These algorithms have convergence guarantees and are spatially distributed with respect to appropriate proximity graphs. Numerical simulations illustrate the results. Mathematics Subject Classification. 37N35, 49J52, 68W15, 93D20. Received January 22, 2004. Revised November 18, 2004. Introduction The current technological development of relatively inexpensive communication, computation, and sensing devices has lead to an intense research activity devoted to the distributed control and coordination of networked systems. In robotic settings, the study of large groups of autonomous vehicles is nowadays a timely concern. The potential advantages of networked robotic systems are their versatility and robustness in the realization of multiple tasks such as manipulation in hazardous environments, pollution detection, estimation and mapbuilding of partially known or unknown environments. A fundamental problem in the distributed coordination of mobile robots is that of providing stable and decentralized control laws that are scalable with the number of agents in the network. Indeed, since the initial works from the robotics and ecology communities on similar problems on swarms and flocking [1, 25, 28], there have been various efforts to provide rigorous procedures with convergence guarantees using a combination of potential energy shaping methods, gyroscopic forces, and graph theory (see [16,22,26,27] and references therein). In our previous work [7, 8], we studied distributed algorithms for deployment and optimal coverage problems using tools from computational geometry, nonsmooth analysis and geometric optimization. The great interest in coordination problems can be easily detected in the proceedings of the 42nd IEEE Conference on Decision and Control held in 2003, the 4th International Conference on Cooperative Control and Optimization held in 2003, or the 6th International Conference on Distributed Autonomous Robotic Systems held in 2002.