Toward Intelligent Navigation for Autonomous Mobile Robots: Learning from the Classics

This chapter provides a comprehensive review of the literature on learning-based approaches to mobile robot navigation. The authors review existing research on end-to-end learning, subsystem replacement, and component adaptation approaches and evaluate each approach based on several criteria such as performance, interpretability, safety, and scalability. They also identify several challenges and future research directions for advancing the field, including improving generalization across different environments and situations, ensuring safety and reliability in real-world applications, integrating learned models with classical methods for hybrid systems, and developing methods for online adaptation and lifelong learning. Overall, this chapter provides a valuable resource for researchers interested in learning-based approaches to mobile robot navigation. By synthesizing the existing literature on this topic and providing a roadmap for future research directions, the authors have contributed to advancing our understanding of how machine learning can be used to improve mobile robot navigation.