Key academic papers shaping the development of humanoid robots — locomotion, manipulation, sim-to-real transfer, VLA models, and tactile sensing.
A system enabling humanoid robots to shadow and imitate human motions in real time using egocentric video, achieving robust whole-body control and skill transfer.
OpenVLA is a 7B-parameter open-source VLA model trained on 970k robot demonstrations, achieving state-of-the-art performance on manipulation benchmarks.
Training bipedal locomotion policies in simulation that transfer zero-shot to real hardware, with training completing in under 20 minutes on a single GPU.
A novel tactile sensor design and simulation framework that enables zero-shot transfer of tactile-guided dexterous manipulation policies to physical robot hands.
GR-2 leverages internet-scale video pretraining to build a generalist manipulation policy that generalizes across robot morphologies and task types.
Training legged robots to perform parkour maneuvers including wall-running, gap jumping, and flipping using a hierarchical RL framework in Isaac Gym.
A hierarchical whole-body control framework that simultaneously manages contact forces, task objectives, and joint limits in real time on 42-DOF humanoid platforms.
A comprehensive survey of sim-to-real transfer techniques applied to humanoid robots, covering domain randomization, system identification, and adaptive control.