Multi-Robot Systems and Distributed Intelligence: The ETHNOS Approach to Heterogeneity
Antonio Sgorbissa
- 发表年份
- 2006
- 引用次数
- 5
- 访问权限
- 开放获取
摘要
In the summer of 1999 the Azzurra Robot Team wins the second prize in the RoboCup competition in Stockholm. In the semi-finals, ART defeats the illustrious CS Freiburg (world champion in Paris 1998, Melbourne 2000, and Seattle 2001) in an unforgettable match that almost causes a heart attack in the supporters of either team. However, in the finals, ART is in difficulty against the very strong and quick team from Sharif CE University, Iran. Sharif players, which have been defeated by ART in the eliminatory rounds, seem to have learnt the lesson. Taking advantage of their undoubted superiority in mechanics, and the robustness of their vision and control algorithms (which, incredibly, run in MSDOS, the OS image being loaded from a bootable disquette), Sharif players are able to easily dribble ART players and to score three times before the end of the match, thus winning the first prize with a final score of 3-1. In contrast with this preamble, RoboCup is not the subject of the discussion here: ART is rather taken as a case study of a "successful" Multi Robot system with very unique characteristics. This is perfectly in accordance with the declared purposes of RoboCup; that is, favouring scientific and technological advancements in robotics and embedded AI with the final aim of transferring the achieved results to real-world applications The choice of robotic soccer as a test field for the design of "autonomous agents" is justified by the intuition that -in most real-world scenarios -a step ahead towards Multi Robot systems is fundamental, if one wants to address issues such as efficiency, fault tolerance, quick response to user requests, or simply consider tasks which cannot be achieved by a single robot. Some examples are autonomous surveillance, rescue operations in large-scale disasters such as earthquakes or fires, humanitarian demining, cooperative manipulation or transportations of objects and furniture, autonomous explorations on desert areas, abandoned mines, or remote planets. To fully take benefit of the larger number of available resources, coordination is important if not essential; the underlying assumption in RoboCup is that these issues can be in part investigated by trying to coordinate a team of soccer-playing robots. In this context, ART uniqueness is not due to the robots' perceptual skills, the mechanisms adopted for behaviour selection and execution, their coordination, localization, or planning capabilities. Its peculiarity is rather the extreme heterogeneity of the team itself: differently from the usual approach, in which a team is designed and built within a single University or
关键词
相关论文
Statistical Learning Theory
Yuhai Wu, Vladimir Vapnik
1999
Artificial intelligence: a modern approach
1995
Applied Nonlinear Control
Jean-Jacques Slotine, Weiping Li
1991
A new optimizer using particle swarm theory
R.C. Eberhart, James Kennedy
2002