Social RoboticsUC San Diego humanoid learns expressive gestures from human dance data
Training on motion capture and dance videos lets a robot wave, high-five and hug while walking.
A UC San Diego team led by Xiaolong Wang trained a humanoid on varied human motion data, including dance videos, training upper and lower body separately before combining them in one control policy. In simulation and then on a real robot, it produced social gestures such as waving, high-fiving and hugging while walking on gravel, grass and slopes. A human operator still steered it with a game controller. The work, presented at Robotics: Science and Systems 2024, targets humanoids that move in ways people find natural to interact with.
Source: ScienceDaily
This archive entry summarises reporting published elsewhere; follow the source link for the full story. The summary was drafted with AI from that reporting.