Whole-Body Robotic Manipulation
Overview
Whole-body robotic manipulation studies how robots coordinate hands and arms with their torso, legs, or mobile base to reach, carry, push, and interact while staying stable. Approaches include inverse kinematics and model-predictive control, reinforcement learning and imitation from human demonstrations, and vision-language-action policies, often combined with tactile or proprioceptive feedback. Work spans humanoids, mobile manipulators, quadrupeds, and aerial robots, with tasks such as household pick-and-place, door opening, object transport, and contact-rich interaction; benchmarks such as BiGym evaluate household skills alongside balance and stepping. A central challenge is making learned policies generalize safely across objects, environments, and long task sequences without requiring large amounts of robot-specific data or sacrificing precise contact control.
Latest papers
88 papers32 papers in the last four weeks, up 113% on the four weeks before. 0.2% of all new papers.