Stochasticity-Induced Uniform Coverage: A Low-Cost Swarm Solution Using Sensors With Limited Field of View

Space coverage is an essential task in many large-scale multi-robot applications, yet existing approaches typically assume ideal sensing and communication capabilities for robots. This paper presents a novel solution for uniform space coverage using robots with limited field of view (FOV), operating without communication infrastructure or global positioning. First, a potential-based coordination strategy is designed where robots react solely to local anisotropic sensory inputs. The stability of the coordination strategy is analyzed theoretically. Second, a distinguishing phenomenon is identified wherein isotropic uniform space coverage patterns emerge from stochastic interactions among anisotropic robots, leading to the development of a stochastic sensor rotation strategy. The effectiveness of the algorithm is validated through distributed experimental implementation on a ground robot group and extensive numerical simulations. The results demonstrate that the proposed algorithm can achieve uniform space coverage with limited FOV sensors, providing a cost-effective solution for large-scale real-world applications. Note to Practitioners - This work addresses a practical challenge in swarm robotics: achieving uniform space coverage with cost-effective robots. Our approach enables uniform coverage using only limited FOV sensors and basic safety sensors, without requiring inter-robot communication or global positioning systems. The implementation involves two key components: a coordination strategy that maintains system stability despite limited sensing, and a stochastic sensor rotation mechanism that enables uniform coverage. The conducted physical experiments utilized depth cameras serving as the coordination sensor and single-line LiDAR as the safety sensor to achieve uniform space coverage. The required sensing activity for coordination and collision avoidance can also be implemented using more cost-effective sensors like monocular cameras and ultrasonic sensors. The proposed algorithm is particularly valuable for large-scale deployments in challenging environments where inter-robot communication is limited or expensive sensors are unavailable.