Humanoid learns to carry objects using video and onboard cameras
From Video to Vision: Learning Generalizable Humanoid Carrying from Human Motion
Robotics
Summary
Humanoid robots usually need special tracking systems and fixed paths to learn how to carry objects, which limits where they can work. The authors developed a new training method called MOVER that learns from videos instead of just motion-capture data. This method allows the robot to carry objects from different places and see the objects using its own camera, without needing an external tracking system. They tested this approach in simulations and on a real robot, showing it can adapt to new object positions more flexibly.
What this means in practice
- •For roboticists: Develop robots that carry objects using onboard vision, without needing external motion capture or fixed paths.
- •For warehouse automation teams: Improve robot object handling in dynamic environments by allowing robots to pick and carry from varying object locations using onboard cameras.
Authors
Ziluo Ding
Abstract
Most existing approaches rely on video or motion-capture data for reinforcement learning (RL) training; however, all of these methods require an upper-level module to generate reference trajectories in real time. In object carrying tasks, the humanoid can only lift objects from designated locations along the reference trajectory, resulting in poor spatial generalization. Moreover, existing policies can only be evaluated inside a motion-capture studio. To address these limitations, we propose \textbf{MOVER}, a multi-stage, Teacher-Assistant-Student, RL training framework for humanoid object carrying. We first collect data from videos and employ GENMO and GMR to retarget the video data, combining this with data augmentation techniques to generalize the data across objects of varying shapes and sizes. In the \textbf{first stage}, building upon the DeepMimic \cite{peng2018deepmimic} motion-training framework, we train the humanoid to transport objects from specified locations along designated trajectories. In the \textbf{second stage}, we relax the tracking of the reference motion and adjust the reward so that the humanoid learns to transport objects from different positions. In the \textbf{third stage}, we apply a Dagger-PPO distillation method to distill away the proprioceptive position observations and reference joint position of the humanoid. This allows the deployed policy to operate without relying on global position information from a motion-capture system; instead, the humanoid can directly perceive the object's pose from its onboard RGB-D camera to complete the carrying task. We design a series of simulation experiments to compare the success rates across stages (Stages 1 vs. 2) and before versus after distillation. We further transfer the policy to a real humanoid, successfully transporting objects at varying positions in a motion-capture-free setting.