Dynamic Assignment in Distributed Motion Planning with Limited Information

Abstract

Abstract-Distributed motion planning of multiple agents raises fundamental and novel problems in control theory and robotics. In particular, in applications such as coverage by mobile sensor networks or multiple target tracking, a great new challenge is the development of motion planning algorithms that dynamically assign targets or destinations to multiple homogeneous agents, not relying on any a priori assignment of agents to destinations. In this paper, we address this challenge using two novel ideas. First, distributed multidestination potential fields are developed that are able to drive every agent to any available destination. Second, nearest neighbor coordination protocols are developed ensuring that distinct agents are assigned to distinct destinations. Integration of the overall system results in a distributed, multiagent, hybrid system for which we show that the mutual exclusion property of the final assignment is guaranteed for almost all initial conditions. Furthermore, we show that our dynamic assignment algorithm will converge after exploring at most a polynomial number of assignments, dramatically reducing the combinatorial nature of purely discrete assignment problems. Our scalable approach is illustrated with nontrivial computer simulations.Index Terms-Distributed control, hybrid systems, motion planning, multiagent assignment problems.

Optimal foraging of renewable resources

“…In other works addressing UAV task-assigment, target locations are known and an assignment strategy is sought that maximizes the global success rate [19], [20] . Moreover, this work holds a place in the recent wave of investigation into the cooperative control of multi-agent systems [21], [22]. Other works addressing cooperative task completion by UAVs include [23], [24], [25].…”

Probabilistic Analysis of Market-based Algorithms for Initial Robotic Formations

“…Several distributed approaches have been developed recently. A distributed heuristic with local communication is explained in (Yun and Rus 2007), while in (Zavlanos and Pappas 2007) the agents are controlled by hybrid models using distributed potential fields. Both approaches fail to return a highly efficient solution since more than one robot can execute the same task for a period of time (until the robots realize they have been executing the same task).…”

“…Therefore, their implementation differs from the classical market-based approach (used in this work) and computes the costs using the information of all of the tasks plus the local information of the robot. There are other recent works on theoretical bounds in Smith and Bullo (2007), Yun and Rus (2007), and Zavlanos and Pappas (2007) but their algorithms are not based on auctions. An important point to note here is that the above-mentioned analyses derive a worst-case bound, a "pessimistic" bound since it is computed assuming that the worst scenario occurs, which may be an unlikely event1 actually, it may not ever happen in a real experiment.…”

A probabilistic model for the performance analysis of a distributed task allocation algorithm

“…There are other recent works on theoretical bounds but their algorithms are not based on auctions. In [11] a distributed heuristic with local communication is explained, while in [12] the agents are controlled by hybrid models using distributed potential fields. Both approaches fail to return a highly efficient solution since more than one robot can execute the same task.…”