Robot Navigation Based on Reinforcement Learning: An Overview

In recent years, the market for real-world robots has become more and more extensive; robots in warehousing and logistics, public services, home management, and other fields are becoming more and more popular. The market demand for better navigation algorithms is becoming stronger, and algorithms based on reinforcement learning are the key focus of future research. In this overview, we analyze the existing three main types of problems in this field and summarize the existing cutting-edge algorithms. Then, we reveal the remaining problems and optimizations in this area and give possible solutions and future research directions.