Multi-Robot Systems

Multi-robot systems (MRS) are becoming one of the most important areas of research in Robotics, due to the challenging nature of the involved research and to the multiple potential applications to areas such as autonomous sensor networks, building surveillance, transportation of large objects, air and underwater pollution monitoring, forest fire detection, transportation systems, or search and rescue after large-scale disasters. Even problems that can be handled by a single multi-skilled robot may
benefit from the alternative usage of a robot team, since robustness and reliability can often be increased by combining several robots which are individually less robust and reliable.



One can find similar examples in human work: several people in line are able to move a bucket, from a water source to a fire, faster and with less individual effort. Also, if one or more of the individuals leaves the team, the task can still be accomplished by the remaining ones, even if slower than before. Another example is the surveillance of a large area by several people. If adequately coordinated,
the team is able to perform the job faster and with reduced cost than a single person carrying out all the work, especially if the cost of moving over large distances is prohibitive. A larger rank of task domains, distributed sensing and action, and insight into social and life sciences are other advantages that can be brought by the study and use of MRS. The relevance of MRS comes also from its inherent inter-disciplinarity.

At the Intelligent Systems Lab of the Institute for Systems and Robotics at Instituto Superior Técnico (ISR/IST), we have been pursuing for several years now an approach to MRS that merges the contributions from two fields: Systems and Control Theory and Distributed Artificial Intelligence. Some of the current problems in the two areas are creating a natural trend towards joint research approaches to their solution. Distributed Artificial Intelligence focuses on multi-agent systems, either virtual (e.g., agents) or with a physical body (e.g., robots), with a special interest on organizational issues, distributed decision making and social relations. Systems and Control Theory faces the growing complexity of the actual systems to be modelled and controlled, as well as the challenges of integrating design, real-time and operation aspects of modern control systems, many of them distributed in nature (e.g., large plant process control, robots, communication networks).


Introduction


Some of the most important, and specific to the area, scientific challenges one can identify in the research on MRS are, to name but the most relevant:

1)The uncertainty in sensing and in the result of actions over the environment inherent to robots, posing serious challenges to the existing methodologies for Multi-Agent Systems (MAS), which rarely take uncertainty into account.

2) The added complexity of the knowledge representation and reasoning, planning, task allocation, scheduling, execution control and learning problems when a distributed setup is considered, i.e., when there are multiple autonomous robots interacting in a common environment, and specially if they have to cooperate in order to achieve their common and individual goals.

3) The noisy and limited bandwidth communications among teammates in a cooperative setting, a scenario which gets worse as the number of team members increase and/or whenever an opponent team using communications in the same range is present.

4) The need to integrate several methodologies that handle the subsystems of each individual robot (extended to the robot team in a cooperative setting) in a consistent manner, such that the integration becomes the most important problem to be solved, ensuring a timely execution of planned tasks.