Interference as a tool for designing and evaluating multi-robot controllers

  • Authors:

  • Dani Goldberg;Maja J. Matarić

  • Affiliations:

  • Computer Science Department, Brandeis University, Waltham, MA;Computer Science Department, Brandeis University, Waltham, MA

  • Venue:
  • AAAI'97/IAAI'97 Proceedings of the fourteenth national conference on artificial intelligence and ninth conference on Innovative applications of artificial intelligence
  • Year:
  • 1997

Quantified Score

Hi-index 0.00

Visualization

Abstract

Designing and implementing cooperative group behaviors for robots is considered something of a black art involving an extensive amount of reprogramming and parameter adjustment. What seems to be lacking is a pragmatic, practical, general-purpose tool that would both guide the design and structure the evaluation of controllers for distributed real-world multi-robot tasks. In this paper, we propose the use of interference. between robots as one such simple tool for designing and evaluating multi-robot controllers. We explore how key issues in multi-robot control can be addressed using interference, a directly measurable property of a multi-robot system. We discuss how behavior arbitration schemes, i.e., the choice of controllers, can be made and adjusted using interference. As an experimental example, we demonstrate three different implementations of a collection clean-up (foraging) task using four physical mobile robots, and present analyses of the experimental data gathered from trials of all three implementations.