Legged Robot

Legged robots aim to create machines capable of robust and agile locomotion across diverse terrains, mimicking the adaptability of animals. Current research heavily focuses on improving state estimation (often using Kalman filters or invariant Kalman filtering), developing robust control policies through reinforcement learning (RL) and model predictive control (MPC), and integrating vision and language models for enhanced perception and task understanding. These advancements are driving progress in applications ranging from industrial inspection to search and rescue, highlighting the potential for legged robots to operate effectively in unstructured and challenging environments.