paper preview
01arXiv
Physics-Guided Biomechanical Gait Adaptation for Humanoid Locomotion on Extreme Sloped Terrains
Teaches a G1 humanoid to walk blind on very steep slopes by aligning balance control to the terrain.
Paper Sweeps
Humanoid Papers from July 5 to July 11, 2026
1Locomotion1Motion Tracking2Loco-Manipulation1Whole-Body Control2Training Infrastructure
1 paper
Locomotion
Terrain, gait, robustness, and agile movement.
1 paper
Motion Tracking
Reference motion, imitation, retargeting, and behavior priors.
project clip
02arXiv
ARDY: Autoregressive Diffusion with Hybrid Representation for Interactive Human Motion Generation
Generates whole-body humanoid motion from text and movement constraints fast enough for interactive robot-control pipelines.
2 papers
Loco-Manipulation
Whole-body task contact: feet, torso, hands, and objects.
paper preview
03arXiv
WristMimic: Full-Body Humanoid Control with Wrist-Guided Manipulation
Teaches humanoids to manipulate objects by guiding the whole body from wrist motion instead of exact finger poses.
project clip
04arXiv
ContactMimic: Humanoid Object Interaction via Contact Control
Teaches humanoids to interact with objects by tracking which body parts should make contact, not just where the body moves.
1 paper
Whole-Body Control
MPC, impedance, safety layers, and whole-body coordination.
paper preview
05arXiv
Immersive Social Interaction with VR and LLM-Assisted Humanoids
Lets people control a humanoid with VR hand tracking and voice locomotion while collecting whole-body interaction data.
2 papers
Training Infrastructure
Benchmarks, datasets, rewards, simulators, and tooling.
paper preview
06arXiv
ThorArena: Benchmarking Humanoid Physical Interaction with Human Motion-Force Demonstrations
Tests whether humanoids can handle contact-rich tasks by replaying human motion and hand-force demonstrations.
paper preview
07arXiv
Behavior Foundations for Quadruped Robots: ABot-C0 Technical Report
Trains quadrupeds to reuse many movement skills from large motion data, then improves them with residual learning.