6 papers
Flow Policy Gradients for Robot Control
Brent Yi, Hongsuk Choi, Himanshu Gaurav Singh +9
Likelihood-based policy gradient methods are the dominant approach for training robot control policies from rewards. These methods rely on differentiable action likelihoods, which…
Learning Sim-to-Real Humanoid Locomotion in 15 Minutes
Younggyo Seo, Carmelo Sferrazza, Juyue Chen +3
Massively parallel simulation has reduced reinforcement learning (RL) training time for robots from days to minutes. However, achieving fast and reliable sim-to-real RL for humanoi…
TWIST2: Scalable, Portable, and Holistic Humanoid Data Collection System
Yanjie Ze, Siheng Zhao, Weizhuo Wang +6
Large-scale data has driven breakthroughs in robotics, from language models to vision-language-action models in bimanual manipulation. However, humanoid robotics lacks equally effe…
GaussGym: An open-source real-to-sim framework for learning locomotion from pixels
Alejandro Escontrela, Justin Kerr, Arthur Allshire +4
We present a novel approach for photorealistic robot simulation that integrates 3D Gaussian Splatting as a drop-in renderer within vectorized physics simulators such as IsaacGym. T…
ResMimic: From General Motion Tracking to Humanoid Whole-body Loco-Manipulation via Residual Learning
Siheng Zhao, Yanjie Ze, Yue Wang +4
Humanoid whole-body loco-manipulation promises transformative capabilities for daily service and warehouse tasks. While recent advances in general motion tracking (GMT) have enable…
Residual Off-Policy RL for Finetuning Behavior Cloning Policies
Lars Ankile, Zhenyu Jiang, Rocky Duan +3
Recent advances in behavior cloning (BC) have enabled impressive visuomotor control policies. However, these approaches are limited by the quality of human demonstrations, the manu…