The authors used model predictive control (MPC) to guide reinforcement learning with predictions of how external forces affect the robot’s centroidal dynamics, meaning the motion of its overall mass and momentum.
HULK teaches humanoids to carry heavy loads while staying balanced
The framework combines model-based guidance, reinforcement learning and training-time balance constraints in a single whole-body control policy.
Big Tech
An Dang · Arturo Flores Alvarez · Yu-Ming Chen · Conor Mc Gartoll · Helen Sun · Aaron Ames · +2 more
University of Michigan · University of California, Los Angeles · Amazon · California Institute of Technology
Research Digest··3 min read
Dang et al.
Why this paper
From Amazon and 3 others
In one line
HULK uses model predictive control to guide reinforcement learning for humanoid whole-body forceful loco-manipulation under sustained payloads.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ✓Limitations stated by the authors (4 noted)
- ✓Reports numbers on named benchmarks (2 benchmarks)
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§