Closed Loop Planning

Closed-loop planning focuses on generating and adapting action sequences for autonomous systems, particularly robots and vehicles, to achieve goals in dynamic, uncertain environments. Current research emphasizes integrating advanced models like large language models (LLMs) and Markov Decision Processes (MDPs) with classical planning techniques, often employing hierarchical or hybrid approaches to balance speed and accuracy. This work aims to improve the robustness, efficiency, and safety of autonomous systems across diverse applications, from robotic manipulation and aerial navigation to autonomous driving, by enabling real-time adaptation to unforeseen events and uncertainties. The resulting improvements in planning capabilities have significant implications for the safety and reliability of these systems in real-world deployments.