Bipedal-Walking-Dynamics Model on Granular Terrains
A new 3-DoF dynamic model tackles the sinkage and slip that make walking on sand so unstable for robots.
A research team from institutions including Xunjie Chen, Xinyan Huang, and Jingang Yi has published a new paper, "Bipedal-Walking-Dynamics Model on Granular Terrains," accepted for ICRA 2026. The work addresses a major challenge in robotics: the instability caused when bipedal robots walk on yielding surfaces like sand or gravel. Traditional models struggle because the foot sinks and slips unpredictably, making it hard to calculate the forces needed for stable walking. The team's breakthrough is a novel dynamics-modeling approach that integrates a dynamic foot-terrain interaction model.
This model adds three additional degrees of freedom (DoF) specifically to estimate foot sinkage and slip—key factors that directly impact a robot's kinematics, kinetics, and energy efficiency, measured as cost of transport (CoT). By accurately computing the ground reaction force (GRF) on granular media, the model can predict media intrusion and gait performance. The researchers validated their model through experiments with a bipedal robotic walker on actual sand, confirming its accuracy in predicting gait profiles and forces. This validated model now serves as a foundational tool for developing advanced control and optimization algorithms, enabling future bipedal robots to walk efficiently and reliably in challenging, unstructured outdoor environments.
- Introduces a 3-DoF dynamic model to predict foot sinkage and slip on granular media like sand.
- Validated with real-world bipedal robot experiments, accurately estimating ground reaction forces (GRF) and cost of transport (CoT).
- Provides a critical tool for optimizing locomotion control, enabling stable and efficient robot walking in unstructured environments.
Why It Matters
This model is a key step toward deploying reliable rescue or exploration robots that can navigate real-world terrain like beaches or disaster zones.