Nav-SCOPE: Swarm Robot Cooperative Perception and Coordinated Navigation
Chenxi Li, Weining Lu, Qingquan Lin, Litong Meng, Haolu Li, Bin Liang
AI summary
Problem
Existing multi-robot navigation methods struggle with high computational and communication demands, limited local perception, and sim-to-real gaps, making scalable coordination in unknown environments difficult.
Approach
Nav-SCOPE rapidly extracts safe directions and key environmental features using FFT-based digital filtering, then shares these compressed byte-level features via an event-triggered ad-hoc network to generate probabilistic interaction fields that guide decentralized path optimization.
Key results
- Reduced path redundancy and improved swarm coordination in simulations
- Robust real-world deployment on ground mobile robots without sim-to-real gaps
- Minimal computation and communication overhead through byte-level feature encoding
- Outperforms state-of-the-art decentralized navigation baselines
Why it matters
Provides a deployable, scalable solution for ground robot swarms to navigate complex environments efficiently without relying on external localization or heavy onboard processing.
Abstract
This paper proposes a lightweight decentralized solution for multi-robot coordinated navigation with cooperative perception. First, we introduce a rapid way to process sensory data, thus obtaining safe directions and key environmental fea- tures. Then, an information flow is created to facilitate real-time perception sharing over wireless ad-hoc networks. Consequently, the environmental uncertainties of each robot are reduced by interaction fields that deliver complementary information. Finally, path optimization is achieved in a probabilistic way, enabling self-organized coordination with effective convergence, divergence, and collision avoidance. Our method is fully inter- pretable and ready for deployment without gaps. Comprehensive simulations and real-world experiments demonstrate reduced path redundancy, robust performance across various tasks, and minimal demands on computation and communication. Note to Practitioners—Local perception is the information source for robot navigation in unknown environments. The extended perception at the swarm level provides complementary information for each robot. This can optimize robot paths and achieve coordinated navigation in a distributed way. However, this purpose is currently unachievable due to the high demands of communication and computation on board. This study reduces these two demands to the extreme by Fast Fourier Trans- form (FFT), digital filtering, probabilistic fusion, and swarm interaction forces. At the same time, our fully interpretable method avoids sim-to-real gaps and can be deployed directly on ground mobile robots. Simulation and real-world experiments outperform a state-of-the-art approach. Future work aims to extend our method to aerial robots as drones.