A New Collision Avoidance Approach for Automated Guided Vehicle Systems Based on Finite State Machines

Abstract

Automated guided vehicles are transportation systems that are widely used in factories, warehouses, and distribution centers. It is of great importance to ensure the control and coordination of vehicles for safe and efficient transportation in multi-vehicle systems. In this study, a control strategy is proposed to enforce collision avoidance of automated guided vehicles operating in a shared zone and overlapping route environment. In the proposed method, while finite state machines are used to model the movement of automated guided vehicles in the environment, the Q-learning method, one of the most common reinforcement learning algorithms, is used for collision avoidance. The presented approach uses the decentralized node-based approach to reduce computational complexity. The proposed method has been validated through simulation performed with vehicle applications that can move both unidirectional and bidirectional. The simulation results show that our presented approach can avoid potential collisions and greatly increase overall efficiency.