Humanoid Loco-Manipulation
Overview
Humanoid loco-manipulation studies how a humanoid can move through its environment while reaching, grasping, pushing, or carrying objects without losing balance. Common approaches combine model-based whole-body control with reinforcement learning or imitation, often using motion retargeting, hierarchical skills, and vision-language or world-model planners. Human demonstrations, simulation-generated data, and robot experience train policies that are evaluated in simulation and on physical humanoids. A central challenge is making these systems reliable across unfamiliar scenes and object loads, especially during contact, while keeping actions physically feasible and enabling recovery from errors.
Latest papers
93 papers31 papers in the last four weeks, up 107% on the four weeks before. 0.2% of all new papers.