Reinforcement Learning Models for Optimizing Emergency Resource Dispatch Strategies: A Review

Abstract

With socioeconomic development, emergency resource dispatch plays a critical role in responding to sudden incidents. As an emerging machine learning technique, reinforcement learning offers new avenues for optimizing emergency resource dispatch strategies. This paper systematically reviews reinforcement learning models for optimizing emergency resource dispatch strategies. It first delves into the theoretical foundations of reinforcement learning, including Markov decision processes and the core concepts of reinforcement learning algorithms. Subsequently, relevant models are categorized and reviewed based on two dimensions: reinforcement learning algorithm types and characteristics of emergency resource dispatch problems. Furthermore, key influencing factors of reinforcement learning models in emergency resource dispatch are analyzed, such as state space design, action space definition, reward function design, and environment modeling. Current challenges and future development directions are identified, aiming to provide references for further research.