Perception-Driven Estimation of Terrain Motion Resistance for UGVs
Tom Bourbon, Stephanie Aravecchia, Cedric Pradalier
AI summary
Problem
Accurate estimation of wheel–terrain interaction parameters is critical for efficient UGV navigation in unstructured environments, yet predicting motion resistance before physical contact remains challenging due to terrain variability and sensor noise.
Approach
The method learns motion resistance distributions from proprioceptive feedback on reference terrains, then transfers this knowledge to new environments by inferring resistance from exteroceptive LiDAR and camera data conditioned on terrain geometry and class using Gaussian-MLP and Gaussian Process Regressor models.
Key results
- Accurate prediction of motion resistance distributions using Gaussian-MLP and GPR models
- Quantitative robustness analysis under domain shifts between training and deployment terrains
- Reliable near-to-far inference using only exteroceptive semantic DEM inputs
- Demonstrated sensitivity of motion resistance to terrain class and local geometry
Why it matters
Enables safer and more efficient autonomous off-road navigation by allowing UGVs to anticipate terrain interaction forces before driving over them.
Abstract
Accurate estimation of wheel–terrain interaction parameters is important for efficient navigation of Unmanned Ground Vehicles in unstructured outdoor environments. In this paper, we propose a hybrid data-driven and model-based method to estimate a priori motion resistance, a terrain-specific parameter representing the force opposing wheel motion, which is largely influenced by terrain class and geometry. The proposed method relies on learning motion resistance from proprioceptive feedback collected on reference terrains. This learned model is then transferred to new environments, where motion resistance is inferred from exteroceptive observation, including LiDAR and cameras, leveraging terrain geometry and class information. To capture uncertainty from terrain roughness and sensor noise, we evaluate two probabilistic mod- els predicting motion resistance distributions: a Gaussian-MLP and a Gaussian Process Regressor. Their robustness to domain shifts is assessed by measuring performance degradation as the target diverges from the source domain. Extensive off-road field experiments validate the method’s effectiveness, demonstrating accurate prediction of motion resistance and its potential for deployment.