Multi-Agent Collision Avoidance Method Using Fuzzy Risk Estimation and Information Sharing in Unknown Environments

Abstract

Automated vehicles within Industry 4.0 are used as logistics units where they can move resources from one place to another safely and efficiently. The automated vehicles can be tasked to work in unknown environments where collision-free navigation is challenging due to uncertainty and lack of environmental information. Collisions can damage equipment and may even cause harm to human workers sharing the same space. In order to carry out tasks and avoid collisions in unknown environments, automated vehicles need means of self regulation and environmental awareness through sensors. This paper presents a multi-agent collision avoidance method using fuzzy risk estimation and information sharing for automated vehicles navigating unknown environments. The automated vehicles exchange information about their current state and GPS location with other automated vehicles and use fuzzy collision risk estimation based on sensor data to avoid collisions with static obstacles and each other. Additionally, the automated vehicles collectively learn about their work environment over time by creating a virtual map of the environment through shared information. To test and evaluate the method, a series of simulated tests are carried out. The results from tests show that the automated vehicles are able to avoid collisions while also accurately mapping an initially unknown work environment using the method.