Dynamic multi-robot team reconfiguration using weighted voting games

Abstract

We consider the problem of dynamic reconfiguration of robot teams when they encounter obstacles while navigating in formation, in an initially unknown environment. We have used a framework from coalition game theory called weighted voting games to analyse this problem and proposed two heuristics that can appropriately partition a robot team into sub-teams. We have experimentally verified our technique on teams of e-puck robots of different sizes and with different obstacle geometries, both on the Webots simulator and on physical robots. We have also shown that our technique performs faster and generates considerably fewer partitions than an existing robot coalition formation algorithm.

Ensemble Methods for Cooperative Robotic Learning

“…Dasgupta et al . use the weighted voting game framework, and a technique called DYN‐PERFORM, to help robotic teams decide whether they will maintain their current formation, or split into smaller teams was capable of getting round a previously unknown obstacle, when they encounter it, while moving in an environment.…”

Multi mobile robot task allocation using multi-objective optimization techniques

“…Dasgupta et al in [21] use the Weighted Voting Game framework, and a technique called DYN-PERFORM, to help Robotic teams decide whether they will maintain their current formation or split into smaller teams, capable of getting round a previously unknown obstacle when they encounter it, while moving in an environment.…”