Search papers, labs, and topics across Lattice.
The paper introduces SOLO, a unified framework that enhances humanoid locomotion over complex terrains by addressing the fragility of perceptive policies through improved terrain reconstruction and temporal credit assignment. By employing a Query Reconstructor (QR) that utilizes Fourier-encoded cell queries, SOLO effectively preserves critical terrain details, leading to significant reductions in height-map errors. Additionally, the implementation of Trajectory-Aware MSE (TA-MSE) Distillation allows for better reward propagation, resulting in a mean traversal success rate of 97.5% on challenging terrains, outperforming existing methods.
SOLO achieves a remarkable 97.5% mean traversal success on complex terrains, showcasing a leap in humanoid locomotion capabilities.
Humans traverse complex terrain over long distances without losing balance, whereas perceptive humanoid policies become fragile as perception and control errors accumulate. We present SOLO, a unified framework addressing two compounding causes of this long-horizon fragility: dense terrain reconstruction smooths action-critical details, and pointwise imitation lacks temporal credit assignment. Its Query Reconstructor (QR) uses Fourier-encoded cell queries to retrieve spatially specific evidence from depth-proprioception tokens, preserving sharp terrain boundaries. Trajectory-Aware MSE (TA-MSE) Distillation adds next-state teacher-student disagreement to the PPO reward, enabling Generalized Advantage Estimation to propagate future disagreement penalties to preceding actions. In simulation, QR reduces height-map L1 error by factors of 3.3-4.0, while TA-MSE surpasses PPO and MSE+PPO in curriculum progression. On stress-test terrains, SOLO achieves 97.5% mean traversal success and 96% stepping-stone success, versus 75.0-75.6% and 0-3% for dense-reconstructor variants. Deployed zero-shot with only a chest-mounted depth camera and proprioception, SOLO completes a continuous 1.5-km outdoor route and an indoor mixed-terrain course. Project page: https://sunpihai-up.github.io/solo/
Microsoft Research