A Multi-Robot Architecture for Continuous Planning and Execution using BDI Agents

Effective coordination in robot systems is essential for enabling teams of robots to collaboratively achieve shared goals in dynamic environments. One key solution in this domain is continuous planning, which allows systems to adapt to failures and correct execution when faced with unforeseen circumstances. This paper proposes a novel multi-robot architecture for continuous planning and execution using Belief-Desire-Intention (BDI) agents. Our approach leverages the BDI framework to deliver adaptive, runtime goal-driven decision-making for heterogeneous robots. Central to the architecture is a coordinator that dynamically forms coalitions, enabling the team to rapidly adjust to environmental changes, unforeseen obstacles, and shifting mission priorities. We evaluate this architecture through experiments in the healthcare domain, demonstrating that our continuous planning approach with BDI agents enables robust adaptation and coordination across heterogeneous teams, even in the face of unexpected events. The validation of our architecture indicates that the success rate remains constant regardless of the failure rate.

Authors

• Carlos Joel Tavares da Silva Universidade de Brasília
• Rafael Melo Santos Universidade Federal da Bahia
• Rafael C. Cardoso University of Aberdeen
• Célia Ghedini Ralha Universidade de Brasília

Keywords

• RobotArchitecture
• ContinuousPlanningandExecution
• BDIAgents