Game theoretic colony control of holonic robots inspecting inner surfaces of pipes


Guciz A., Erkmen A. , Erkmen I.

4th International Conference and Exposition/Demonstration on Robotics for Challenging Situations and Environments, New-Mexico, Amerika Birleşik Devletleri, 27 Şubat - 02 Mart 2000, ss.234-240 identifier identifier

  • Doi Numarası: 10.1061/40476(299)30
  • Basıldığı Şehir: New-Mexico
  • Basıldığı Ülke: Amerika Birleşik Devletleri
  • Sayfa Sayıları: ss.234-240

Özet

The purpose of this work is to plan and control task oriented configuration changes in a metamorphic robotic system, formed by a colony of axisymmetric holons moving to inspect pipe inner surfaces. The metamorphic robotic system in question, is a reconfigurable robotic structure made up of identical, independently controlled, mechatronic modules (holons) constituting a colony. The holons in that colony have very limited individual task abilities and are only capable of connecting, disconnecting and rotating about their center of symmetry while climbing over other holons. But it is only when they come together to form a colony that they can be used as a robotic system to perform many tasks, such as pipe surface inspection. The metamorphic inspection system has the advantages of robust reconfigurability, that can only be achieved by an efficient adaptable control mechanism which results from the difficult motion-planning problem for the colony. In this work, we consider three well-known motion-planning strategies that make use of different heuristics: "The Short Path Algorithm, The Artificial Potential Algorithm and The Sub-Goals Algorithm", and in order to generate an adaptable motion-planning and control strategy in possibly changing environments and achieve the best result in any environment, out work proposes to use all three strategies together, combined with the principles of the "mixed strategy" of Game Theory. The general path-planning strategy that we develop in this paper, switches from one of the three strategies to another, so that, at every step, the most suitable strategy is used and a result better than or equal to the "best" of the three strategies is achieved in all cases.