Reproduce “Adaptive Force-Based Control of Dynamic Legged Locomotion over Uneven Terrain” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Mohsen Sombolestan, Quan Nguyen
A novel adaptive force-based control framework integrates L1 adaptive control with Model Predictive Control, enabling quadruped robots to perform dynamic locomotion, such as trotting and bounding, robustly over uneven terrain while carrying substantial unknown loads, without requiring prior environmental models or controller recalibration.
Key insights
- L1 adaptive control effectively decouples adaptation and robustness, providing fast and smooth compensation for significant, continuous model uncertainties and unmodeled environmental interactions in dynamic robotic systems.
- Using an MPC as the reference model within the L1 adaptive framework is crucial for capturing and adapting to complex Center of Mass dynamics during highly agile and dynamic locomotion.
- Integrating the adaptive compensation as a residual vector into an extended MPC state offers a computationally efficient method for combining advanced adaptive control with optimization-based force control.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.