Autonomous mobile robots (AMRs) are increasingly used in various applications, including transportation, logistics, and healthcare. One of the central challenges in AMR deployment is path planning, which involves finding an optimal and collision-free route from the robot’s initial position to its destination. This review, which builds upon a two-part study, presents a comprehensive overview of state-of-the-art techniques for AMR path planning. This paper focuses on classical and heuristic-based strategies, providing valuable insights into their foundational roles in autonomous mobile robot navigation. We categorize existing approaches based on their core principles, including graph-based, heuristic-based, metaheuristic-based, and learning/reasoning-based methods. For each category, we analyze the strengths and limitations of the representative techniques and evaluate their effectiveness in various operational scenarios. Furthermore, we identify emerging research directions and ongoing challenges in AMR path planning, such as multi-robot coordination, adaptation to dynamic environments, and integration with human–robot interaction frameworks.

The rapid growth of technology has introduced robots into daily life, necessitating navigation frameworks that enable safe, human-friendly movement while accounting for social aspects. Such methods must also scale to situations with multiple humans and robots moving simultaneously. Recent advances in Deep Reinforcement Learning (DRL) have enabled policies that incorporate these norms into navigation. This work presents a socially aware navigation framework for mobile robots operating in environments shared with humans and other robots. The approach, based on single-agent DRL, models all interaction types between the ego robot, humans, and other robots. Training uses a reward function balancing task completion, collision avoidance, and maintaining comfortable distances from humans. An attention mechanism enables the framework to extract knowledge about the relative importance of surrounding agents, guiding safer and more efficient navigation. Our approach is tested in both dynamic and static obstacle environments. To improve training efficiency and promote socially appropriate behaviors, Imitation Learning is employed. Comparative evaluations with state-of-the-art methods highlight the advantages of our approach, especially in enhancing safety by reducing collisions and preserving comfort distances. Results confirm the effectiveness of our learned policy and its ability to extract socially relevant knowledge in human–robot environments where social compliance is essential for deployment.

The rapid growth of technology has introduced robots into daily life, necessitating navigation frameworks that enable safe, human-friendly movement while accounting for social aspects. Such methods must also scale to situations with multiple humans and robots moving simultaneously. Recent advances in Deep Reinforcement Learning (DRL) have enabled policies that incorporate these norms into navigation. This work presents a socially aware navigation framework for mobile robots operating in environments shared with humans and other robots. The approach, based on single-agent DRL, models all interaction types between the ego robot, humans, and other robots. Training uses a reward function balancing task completion, collision avoidance, and maintaining comfortable distances from humans. An attention mechanism enables the framework to extract knowledge about the relative importance of surrounding agents, guiding safer and more efficient navigation. Our approach is tested in both dynamic and static obstacle environments. To improve training efficiency and promote socially appropriate behaviors, Imitation Learning is employed. Comparative evaluations with state-of-the-art methods highlight the advantages of our approach, especially in enhancing safety by reducing collisions and preserving comfort distances. Results confirm the effectiveness of our learned policy and its ability to extract socially relevant knowledge in human-robot environments where social compliance is essential for deployment.

Robotic ball balancing is a nonlinear and inherently unstable manipulation taskthat serves as a challenging benchmark for advanced control methods. However,learning-based solutions that are both safe and adaptive remain difficult due to theunderactuated ball–plate dynamics, sparse task success, and the risk of hardwaredamage. In this work, we study a ball-on-plate balancing task on a 7-DoF KINOVAmanipulator modeled in Simscape Multibody and develop a reinforcement learningframework that explicitly targets safety, robustness, and generalization. We firsttrain a Soft Actor-Critic (SAC) agent as a baseline controller, using continuous jointtorques and task-specific reward functions to stabilize the ball around a desiredposition. We then introduce a control barrier-inspired safety term in the rewardto discourage violations of position and tilt limits, and apply transfer learning toadapt the nominal balancing policy to more dynamic ‘catch’ scenarios involvingvertical ball drops. Finally, we evaluate zero-shot generalization to variations in ballmass, radius, and initial position, and compare SAC with DDPG, TD3, and PPOunder the same environment and reward structure. Our analysis shows that theproposed SAC-based safe RL framework achieves reliable balancing, significantlyreduces interactions with unsafe boundaries, and maintains performance across awide range of conditions, highlighting the potential of RL for safe and adaptiverobotic manipulation.