Virtual Traffic Lights for Multi-Robot Navigation: Decentralized Planning with Centralized Conflict Resolution

TL;DR

The paper tackles the challenge of coordinating multiple robots in shared spaces without collisions by blending decentralized path planning with centralized conflict resolution, effectively acting as a virtual traffic light. The central mediator detects conflicts, clusters interacting robots, and applies a priority-based stop-resume policy so only non-conflicting robots proceed through intersections. Key contributions include a platform-agnostic hybrid coordination framework, a 1000-simulation validation showing higher success rates and fewer replans than a purely decentralized approach, and real-world demonstrations on both dynamic quadruped and lane-following wheeled platforms. The work demonstrates scalable, robust coordination suitable for real-world deployment and lays groundwork for extending the approach to 3D environments and larger-scale autonomous traffic management.

Abstract

We present a hybrid multi-robot coordination framework that combines decentralized path planning with centralized conflict resolution. In our approach, each robot autonomously plans its path and shares this information with a centralized node. The centralized system detects potential conflicts and allows only one of the conflicting robots to proceed at a time, instructing others to stop outside the conflicting area to avoid deadlocks. Unlike traditional centralized planning methods, our system does not dictate robot paths but instead provides stop commands, functioning as a virtual traffic light. In simulation experiments with multiple robots, our approach increased the success rate of robots reaching their goals while reducing deadlocks. Furthermore, we successfully validated the system in real-world experiments with two quadruped robots and separately with wheeled Duckiebots.

Figures (7)

• Figure 1: Our hybrid system combines decentralized planning with centralized conflict resolution. To prevent a deadlock, a central node issues a virtual red light to the quadruped on the left, allowing the other robot to pass safely through the conflict zone. Dotted lines represent each robot's traversed path.
• Figure 2: System overview with N robots, where only robots $R_{1}$ and $R_{2}$ have a conflicting path. The centralized coordination system allows $R_{1}$ to proceed and halts $R_{2}$ until $R_{1}$ has navigated through the conflicting intersection.
• Figure 3: Snapshots from a simulation run with 6 robots using the hybrid coordination system. There is a 4x4 grid of pillars depicted in black. Robots are depicted as colored circles with a white line representing their orientation. The colored dotted lines depict each robot's global path to their goals, represented by a star of the same color. Intersections are opaque zones which are green when empty and red when occupied.
• Figure 4: Average success rate (percentage of robots reaching their goal within a 135s timeout) for the Hybrid and Decentralized systems versus the number of robots. Each point is the mean over 500 trials; shaded regions represent the 95% confidence interval.
• Figure 5: Average speed for Hybrid and Decentralized systems versus the number of robots. Speed is measured in (pixels/step) $\times$ 100, averaged from the start of a run until a robot reaches its goal or times out. Each point is the mean over 500 trials; shaded regions represent the 95% confidence interval.