A Framework for Information Distribution, Task Execution and Decision Making in Multi-Robot Systems

Matthias Rambow, Florian Rohrmüller, Omiros Kourakos, Drazen Brscic, Dirk Wollherr, Sandra Hirche, Martin BussIEICE Trans. Information and Systems 2010 · DOI

Abstract

Robotic systems operating in the real-world have to cope with unforeseen events by determining appropriate decisions based on noisy or partial knowledge. In this respect high functional robots are equipped with many sensors and actuators and run multiple processing modules in parallel. The resulting complexity is even further increased in case of cooperative multi-robot systems, since mechanisms for joint operation are needed. In this paper a complete and modular framework that handles this complexity in multi-robot systems is presented. It provides efficient exchange of generated data as well as a generic scheme for task execution and robot coordination.

Introduction

Operating autonomous robots in dynamic environments requires managing high-frequency sensor-actuator loops alongside complex high-level reasoning. In multi-robot systems (MRS), this challenge is compounded by the need for coordination. Traditional approaches often oscillate between centralized control, which is efficient but fragile, and fully distributed control, which is robust but suffers from high communication overhead. This paper proposes a hybrid framework that balances these needs by centralizing intra-robot processing while distributing inter-robot decision-making.

System Architecture

The proposed architecture rests on three primary components: the ARCADE middleware, hierarchical task trees, and a market-based allocation system.

The ARCADE Middleware

The communication backbone, ARCADE, utilizes a dual-speed approach to handle different data requirements. For high-frequency internal communication, it employs the KogMo-RTDB (Real-Time Database). For inter-robot communication, it uses ZeroC/ICE, which facilitates the exchange of abstracted, symbolic information rather than raw sensor data, significantly reducing bandwidth requirements.

The ARCADE middleware architecture showing the integration of real-time databases and inter-robot communication.

Task Execution and Market-Based Allocation

Tasks are structured as hierarchical trees. At the robot level, an "Arbitrator" manages conflicts between subtasks using priority-based logic. At the team level, tasks are distributed via a market-based mechanism. An "Auctioneer" announces tasks, and "Brokers" on individual robots submit bids based on their specific hardware capabilities and current state. This negotiation is supported by Bayesian Networks that predict the likelihood of mission success based on local performance metrics.

Overview of the market-based task allocation system and its interaction with the world model.

Experimental Evaluation

The framework was tested using two heterogeneous mobile manipulators, R1 and R3, in a service scenario. R1 featured a dexterous three-fingered BarrettHand, while R3 utilized a simpler two-finger gripper. The robots were tasked with fulfilling a human's drink order, requiring them to perceive the environment via a 40-camera overhead cluster and coordinate a physical handover of the object.

The results demonstrated that the market-based system naturally encouraged specialization: R1, with its superior manipulation hardware, consistently won bids for complex grasping, while R3 handled social interaction and navigation. The system successfully managed dynamic coalition formation and seamless robot-to-robot handovers without hard-coded scheduling.

Conclusion and Future Work

This framework demonstrates that abstraction and hybrid control are essential for scaling multi-robot systems. By separating high-frequency local control from high-level global negotiation, the system achieves both real-time responsiveness and robust team coordination. Future research will focus on enhancing communication reliability in degraded network environments and deepening the analysis of system interdependencies to improve fault tolerance.

Summary updated

BibTeX
@article{rambow2010framework,
  author = {Matthias Rambow and Florian Rohrmüller and Omiros Kourakos and Drazen Brscic and Dirk Wollherr and Sandra Hirche and Martin Buss},
  title = {{A Framework for Information Distribution, Task Execution and Decision Making in Multi-Robot Systems}},
  journal = {IEICE Trans. Information and Systems},
  year = {2010},
  doi = {10.1587/transinf.e93.d.1352},
}
Figure