A comprehensive survey on SLAM and machine learning approaches for indoor autonomous navigation of mobile robots

The complex and dynamic nature of indoor environments presents significant challenges for mobile robots autonomous navigation. Traditional navigation methods, reliant on handcrafted features and algorithms, often struggle to adapt to these challenges. Recently, machine learning techniques have emerged as a promising approach for indoor autonomous navigation, offering the ability to learn from data to extract features and develop robust navigation strategies. This survey presents recent strategies for indoor autonomous navigation of mobile robots, providing a comprehensive overview of traditional methods for indoor autonomous mobile robots simultaneous localization and mapping mapping (SLAM), path planning and obstacle avoidance, and machine learning approaches, including deep learning and reinforcement learning. Furthermore, the paper discusses the specific challenges of indoor autonomous navigation for mobile robots and examines the advantages, challenges, and limitations of applying machine learning techniques in this context. Since performance evaluation is crucial for proving the efficiency of each novel developed algorithm and method, the most important performance evaluation metrics are described and mathematically presented with formulas. A systematic review on recent advances in indoor autonomous mobile robot navigation is further supported by presenting relevant patents on the topic of the paper and the field. Additionally, the survey identifies promising future research directions in machine learning-based indoor autonomous navigation. Last but not least, this survey aims to serve as a valuable resource for researchers and engineers interested in developing advanced and machine learning-based indoor autonomous navigation systems for mobile robots.