Dynamic Task Allocation in Multi-Robot System Based on a Team Competition Model

Area of Science:

• Robotics
• Artificial Intelligence
• Game Theory

Background:

• Multi-robot systems face dynamic task allocation challenges.
• Integrating game theory offers novel solutions for complex coordination problems.

Purpose of the Study:

• To propose a team-competition game model for dynamic multi-robot task allocation.
• To analyze sub-game perfect strategies for optimal task assignment.

Main Methods:

• Formulated a multi-round team competition model as an extensive-form game.
• Analyzed sub-game perfect equilibria for simultaneous and turn-based team selections.
• Investigated scenarios with and without redundant players and transitive/non-transitive strengths.

Main Results:

• Uniformly random strategy is a sub-game perfect equilibrium when no redundant players exist.
• Teams can discard weak players if player strengths are transitive.
• Counterintuitive results emerge with redundant players and non-transitive strengths, where dominated players can improve team payoff.

Conclusions:

• The team-competition model provides a robust framework for dynamic multi-robot task allocation.
• Game theory uncovers complex, non-intuitive optimal strategies in robot team coordination.
• Understanding player interactions is crucial for maximizing team performance in dynamic environments.