Multi-robot maze exploration using an efficient cost-utility method

TL;DR

This work tackles fully exploring unknown mazes with multiple robots by introducing a cost-utility exploration algorithm that augments the nearest-frontier strategy with a dual utility term computed along the wavefront path. The utility combines a frontier-distance component $u_{mnm}$ and a path-exploration component $u_{jgr}$ through $utility(f) = N(u_{mnm}(f)) + \lambda \cdot N(u_{jgr}(f))$, with normalization $N$ and a tunable parameter $\lambda$ (optimal around $0.2$). The method assigns distinct goals to agents to reduce collisions, includes collision avoidance, and outperforms several baselines (e.g., HEDAC, Flood-Fill, CU variants) across rounds, cost, and efficiency in extensive randomized-maze experiments analyzed with Copeland voting. These results suggest practical advantages for scalable, multi-robot exploration, with potential extensions to larger or dynamic environments and real-world constraints such as communication range and energy considerations.

Abstract

In the field of modern robotics, robots are proving to be useful in tackling high-risk situations, such as navigating hazardous environments like burning buildings, earthquake-stricken areas, or patrolling crime-ridden streets, as well as exploring uncharted caves. These scenarios share similarities with maze exploration problems in terms of complexity. While several methods have been proposed for single-agent systems, ranging from potential fields to flood-fill methods, recent research endeavors have focused on creating methods tailored for multiple agents to enhance the quality and efficiency of maze coverage. The contribution of this paper is the implementation of established maze exploration methods and their comparison with a new cost-utility algorithm designed for multiple agents, which combines the existing methodologies to optimize exploration outcomes. Through a comprehensive and comparative analysis, this paper evaluates the performance of the new approach against the implemented baseline methods from the literature, highlighting its efficacy and potential advantages in various scenarios. The code and experimental results supporting this study are available in the following repository (https://github.com/manouslinard/multiagent-exploration/).

Figures (15)

• Figure 1: Abstract Schema of Socratic Approach.
• Figure 2: Example of potential fields usage in maze exploration.
• Figure 3: Maze converted to Graph (Tree & Graph Approach).
• Figure 4: Visualization of the exploration process. The black cells represent the agents. The light gray cells denote free space, and the darker gray cells represent obstacles.
• Figure 5: Comparison of Copeland scores for different $\lambda$ values across 100 experiments, in maze dimensions 15x15, 30x30, and 50x50.