Frontier Shepherding: A Bio-inspired Multi-robot Framework for Large-Scale Exploration

TL;DR

FroShe addresses large-scale unknown-environment exploration by reframing frontier discovery as a bio-inspired shepherding problem. It introduces a modular, decentralized pipeline with Frontier Processor, Swarm Processor, Predator Processor, and an Exploration Rate Monitor to coordinate multiple robots with minimal tuning. Through ROS Noetic simulations and forest-like real-world tests, FroShe demonstrates robust scalability and up to ~25% faster exploration with three UAVs, along with reduced performance variance across different environments. The approach holds promise for practical autonomous exploration tasks in environmental monitoring and search-and-rescue scenarios.

Abstract

Efficient exploration of large-scale environments remains a critical challenge in robotics, with applications ranging from environmental monitoring to search and rescue operations. This article proposes Frontier Shepherding (FroShe), a bio-inspired multi-robot framework for large-scale exploration. The framework heuristically models frontier exploration based on the shepherding behavior of herding dogs, where frontiers are treated as a swarm of sheep reacting to robots modeled as shepherding dogs. FroShe is robust across varying environment sizes and obstacle densities, requiring minimal parameter tuning for deployment across multiple agents. Simulation results demonstrate that the proposed method performs consistently, regardless of environment complexity, and outperforms state-of-the-art exploration strategies by an average of 20% with three UAVs. The approach was further validated in real-world experiments using single- and dual-drone deployments in a forest-like environment.

Figures (10)

• Figure 1: The FroShe framework. The key contributions of the proposed method are highlighted in green.
• Figure 2: Virtual sheep are represented by blobs, with the radius depicting the weight of each virtual sheep. The sheep's weight, abstractly defining the exploration gain, is determined by the number of unexplored cells in the red grid centered at each sheep. Consequentially, heavier sheep are seen at the corners.
• Figure 3: Collecting mode. (a) Planned trajectory to $P_c$ (b) Planned trajectory from $P_c$ to $C_m$ (c) Final $\mathcal{F}$ in $\mathcal{M}$ after collecting.
• Figure 4: Herding mode (a) Planned trajectory to $P_d$ (b) Planned trajectory from $P_d$ to $C_h$ (c) Final $\mathcal{F}$ in $\mathcal{M}$ after herding.
• Figure 5: Exploration rate monitoring in a single-agent exploration scenario within a $6400m^2$ forest-like environment. The plot tracks the percentage of explored area $E$ over time and its rate of change $\Delta E$. Black dots indicate mode switches triggered by changes in $d_t$, while yellow dots mark instances where the current mode is retained. The Fast Moving Average (FMA) and Slow Moving Average (SMA) guide these transitions, ensuring steady exploration progress.