Collective robotics: from social insects to robots
Adaptive Behavior
Communication in reactive multiagent robotic systems
Autonomous Robots
Lazy learning
A teaching strategy for memory-based control
Lazy learning
Reinforcement Learning
Cooperative Mobile Robotics: Antecedents and Directions
Autonomous Robots
Reinforcement Learning in the Multi-Robot Domain
Autonomous Robots
A Modular Approach to Multi-Agent Reinforcement Learning
ECAI '96 Selected papers from the Workshop on Distributed Artificial Intelligence Meets Machine Learning, Learning in Multi-Agent Environments
The effect of action recognition and robot awareness in cooperative robotic teams
IROS '95 Proceedings of the International Conference on Intelligent Robots and Systems-Volume 1 - Volume 1
Reinforcement learning: a survey
Journal of Artificial Intelligence Research
Reactivity and Deliberation: A Survey on Multi-Robot Systems
Balancing Reactivity and Social Deliberation in Multi-Agent Systems, From RoboCup to Real-World Applications (selected papers from the ECAI 2000 Workshop and additional contributions)
Tunably decentralized algorithms for cooperative target observation
Proceedings of the fourth international joint conference on Autonomous agents and multiagent systems
| Hi-index | 0.00 |
Most of the straight-forward learning approaches in cooperativerobotics imply for each learning robot a state space growthexponential in the number of team members. To remedy theexponentially large state space, we propose to investigate a lessdemanding cooperation mechanism—i.e., various levels ofawareness—instead of communication. We define awareness asthe perception of other robots locations and actions. We recognizefour different levels (or degrees) of awareness which imply differentamounts of additional information and therefore have differentimpacts on the search space size (&THgr;(0), &THgr;(1),&THgr;(N), o(N),^1 where Nis the number of robots in the team). There are trivial arguments in favor ofavoiding binding the increase of the search space size to the numberof team members. We advocate that, by studying the maximum number ofneighbor robots in the application context, it is possible to tunethe parameters associated with a &THgr;(1) increase of the searchspace size and allow good learning performance. We use thecooperative multi-robot observation of multiple moving targets(CMOMMT) application to illustrate our method. We verify thatawareness allows cooperation, that cooperation shows betterperformance than a purely collective behavior and that learnedcooperation shows better results than learned collective behavior.