Adaptive Motion Control of Legged Robots via Predictive Force Modeling

Legged robots are well-suited for navigating complex and uneven terrains, but maintaining stable locomotion in such environments remains a critical challenge. Vision-based systems like cameras and LiDAR often face issues such as noise, occlusion, and latency, which can lead to misinterpretation of terrain and instability during movement. Drawing inspiration from human tactile feedback while walking, this project explores the use of force-based feedback to improve adaptability in legged robots. It introduces a control strategy that combines predictive force modeling with real-time feedback to dynamically adjust foot placement. By anticipating force patterns based on terrain inclination and detecting deviations during locomotion, the system aims to enhance stability without relying heavily on visual perception.