Generic, scalable and decentralized fault detection for robot swarms

Area of Science:

• Robotics
• Artificial Intelligence
• Distributed Systems

Background:

• Robot swarms utilize decentralized control for scalability and robustness.
• Traditional multi-robot systems often lack inherent fault tolerance.
• Existing swarm systems may not actively detect or accommodate faults.

Purpose of the Study:

• To develop a generic fault-detection system for robot swarms.
• To enable robots with limited sensing to identify and classify abnormal behavior.
• To enhance the robustness and fault tolerance of decentralized multi-robot systems.

Main Methods:

• Implementation of an immune system-inspired algorithm for online learning.
• Classification of robot behavior as normal or abnormal based on local perception.
• Systematic performance assessment through simulation experiments.

Main Results:

• The developed fault-detection system demonstrates robustness and timely fault detection.
• The system achieves a low false positive rate in identifying faulty robots.
• Performance was analyzed in a foraging task and under environmental perturbations.

Conclusions:

• The generic fault-detection system significantly improves the fault tolerance of robot swarms.
• This approach enables long-term autonomy for robust multi-robot systems.
• The system has potential applications in distributed intelligent automation.