Traffic control for a swarm of robots: Avoiding group conflicts

Leandro Soriano Marcolino, Luiz ChaimowiczIROS 2009 · OpenReview

Abstract

In swarm robotics, decentralized navigation often leads to congestion and gridlock when multiple groups encounter each other in unstructured environments. This paper introduces a novel coordination algorithm that leverages local communication to prevent such conflicts. By enabling robots at the swarm's periphery to detect oncoming groups and "warn" their teammates, the entire collective can proactively deviate from its path. Experimental results in both simulation and physical hardware demonstrate that this approach can improve navigation efficiency by up to 66% compared to standard local repulsion methods.

Introduction

A fundamental challenge in swarm robotics is maintaining smooth collective motion in the face of congestion. While simple rules like local repulsion and goal-seeking are effective for isolated groups, they frequently fail when two or more swarms meet head-on. In these scenarios, robots often become trapped in a chaotic "clump," where individual avoidance maneuvers cancel each other out, halting forward progress. Traditional solutions typically rely on centralized controllers or rigid environmental structures like lanes, which are incompatible with the flexible, distributed nature of robotic swarms. This research addresses the need for a purely decentralized mechanism to resolve group-level conflicts.

Methodology

The proposed algorithm extends the standard Artificial Potential Field (APF) method by adding a distributed coordination layer. In the APF framework, robots are driven by attractive forces toward a goal and repulsive forces away from obstacles and neighbors. The authors' coordination layer introduces four key components:

  1. Congestion Detection: Robots monitor neighbors within a sensing radius $\delta$. A "congestion risk" is identified if a robot detects another agent moving toward a significantly different target.
  2. Teammate Warning: Upon detecting a risk, the robot broadcasts a local signal to its "teammates" (robots sharing the same target direction) within a communication range $\alpha$.
  3. Finite State Machine (FSM): Robots switch between a Normal Mode (standard APF) and a Deviating Mode. When in Deviating Mode, robots modify their control law to steer at a fixed angle (e.g., counter-clockwise) relative to their target.
  4. Information Relay: To ensure group-wide coordination, teammates who receive a warning relay it to others further back in the swarm. This allows the rear of the group to begin deviating before they directly encounter the oncoming swarm.
The four stages of the coordination algorithm: (a) initial detection and warning, (b) information relay, (c) group-wide deviation, and (d) successful passing.

This mechanism facilitates spontaneous symmetry breaking, allowing the swarms to "flow" around each other similarly to how human crowds naturally form lanes in hallways.

Experimental Evaluation

The algorithm was evaluated through both high-fidelity simulations and physical experiments. Simulations were conducted using the Player/Stage framework, modeling up to 48 P2DX robots in two-group and four-group crossing scenarios.

To validate the real-world feasibility of the approach, the authors utilized a fleet of 12 e-puck robots. These small, differential-drive robots are equipped with infrared sensors for local obstacle detection and Bluetooth for inter-robot communication.

The fleet of 12 e-puck robots used for physical validation of the traffic control algorithm.

Results and Discussion

The coordination algorithm demonstrated significant performance improvements over baseline local repulsion:

  • Efficiency: Navigation time was reduced by 55% in two-group scenarios and 66% in four-group scenarios.
  • Scalability: Unlike the uncoordinated approach, where completion time increased sharply with swarm size, the proposed algorithm maintained nearly constant performance as the number of robots per group grew.
  • Robustness: Physical experiments with e-puck robots confirmed that the decentralized logic is resilient to sensor noise and communication delays inherent in real-world hardware.

The "aha moment" of this research is the realization that local sensing alone is insufficient for efficient swarm navigation. By treating the swarm as a communicative social unit, "scout" robots can provide the necessary lead time for the entire group to act cohesively.

Conclusion

This work provides a scalable, decentralized solution for traffic control in robotic swarms. By hardcoding simple directional deviation rules and enabling local information relay, the system avoids the need for complex negotiation or centralized oversight. Future research will focus on generalizing the algorithm for groups meeting at arbitrary angles and navigating through environments with complex static obstacles.

Summary updated

BibTeX
@inproceedings{marcolino2009traffic,
  author = {Leandro Soriano Marcolino and Luiz Chaimowicz},
  title = {{Traffic control for a swarm of robots: Avoiding group conflicts}},
  booktitle = {IEEE/RSJ International Conference on Intelligent Robots and Systems},
  year = {2009},
  url = {https://openreview.net/forum?id=wFIRPLI8Htu},
}
Figure