This paper presents a comprehensive survey of recent machine learning (ML) approaches applied to mobile robot control, categorizing them into supervised learning, unsupervised learning, and reinforcement learning. The study compares the strengths and weaknesses of these methods and discusses their applications in tasks such as position estimation, environment mapping, SLAM, terrain classification, obstacle avoidance, path following, learning to walk, and multirobot coordination. The paper also distinguishes between wheeled and walking robots and highlights challenges such as computational limitations, real-time decision-making, adaptability to changing environments, data acquisition, and safety. It emphasizes the need for further research in developing ML algorithms for nature-inspired walking robots, as such solutions are rare in the literature. The survey shows that ML methods are widely used in mobile robotics for various tasks, including navigation, control, and remote sensing. Supervised learning methods, such as regression and classification, are applied for tasks like robot localization, obstacle detection, and terrain classification. Unsupervised learning methods, including clustering and dimensionality reduction, are used for SLAM and remote sensing. Reinforcement learning methods are applied for obstacle avoidance and path planning. The paper concludes that ML techniques are essential for advancing mobile robotics, but challenges remain in improving performance, adaptability, and safety.This paper presents a comprehensive survey of recent machine learning (ML) approaches applied to mobile robot control, categorizing them into supervised learning, unsupervised learning, and reinforcement learning. The study compares the strengths and weaknesses of these methods and discusses their applications in tasks such as position estimation, environment mapping, SLAM, terrain classification, obstacle avoidance, path following, learning to walk, and multirobot coordination. The paper also distinguishes between wheeled and walking robots and highlights challenges such as computational limitations, real-time decision-making, adaptability to changing environments, data acquisition, and safety. It emphasizes the need for further research in developing ML algorithms for nature-inspired walking robots, as such solutions are rare in the literature. The survey shows that ML methods are widely used in mobile robotics for various tasks, including navigation, control, and remote sensing. Supervised learning methods, such as regression and classification, are applied for tasks like robot localization, obstacle detection, and terrain classification. Unsupervised learning methods, including clustering and dimensionality reduction, are used for SLAM and remote sensing. Reinforcement learning methods are applied for obstacle avoidance and path planning. The paper concludes that ML techniques are essential for advancing mobile robotics, but challenges remain in improving performance, adaptability, and safety.