UniT has been accepted to RA-L😀! Many thanks to the editors and reviewers for their valuable feedback. Updated Paper: arxiv.org/abs/2408.06481 Code: github.com/ZhengtongXu/Un… Project Website: zhengtongxu.github.io/unit-website/

🤖Introducing TWIST: Teleoperated Whole-Body Imitation System. We develop a humanoid teleoperation system to enable coordinated, versatile, whole-body movements, using a single neural network. This is our first step toward general-purpose robots. 🌐humanoid-teleop.github.io

🦾How can humanoids unlock real strength for heavy-duty loco-manipulation? Meet FALCON🦅: Learning Force-Adaptive Humanoid Loco-Manipulation. 🌐: lecar-lab.github.io/falcon-humanoi… See the details below👇:

🤖Optimization and imitation learning are both powerful tools in robotics, yet they often belong to distinct paradigms and areas. How can we integrate trajectory optimization into visuomotor policies to improve the quality of trajectories? 💡We introduce DiffOG, a recipe for

How can we effectively leverage human videos for robot learning by bridging the inherent embodiment gap? We introduce UniSkill, a universal skill representation, a scalable method for learning cross-embodiment skill representations from large-scale in-the-wild video data. 1/n

💡We introduce Canonical Policy, a 3D equivariant policy that works effectively with a single camera and remains robust even under dynamic camera perturbations. See thread (1/N)👇

This is Zhiyuan’s @zhangzh56938583 first PhD project! Be sure to give him a follow! In the near future, he’ll be sharing more of his exciting work on 3D vision, robot learning, and tactile sensing! It’s been a true pleasure collaborating with Zhiyuan. He’s an exceptionally

🤖Visuomotor policies often fail to generalize to new object poses, appearances, and viewpoints. 💡We introduce Canonical Policy, a 3D equivariant policy that works effectively with a single camera and remains robust even under dynamic camera perturbations. See thread (1/N)👇

😃Very excited to attend ICRA 2025 next week and connect with researchers from around the world! I'll be presenting 3 papers at the conference: 1. Learning Model Predictive Control for Tactile-reactive Grasping (transferred from T-RO) 📍Oral presentation at 8:30 am - 8:35 am,

TWIST now supports Dex hand teleop. And delay is half. We are still actively improving our system. Human-level whole-body manipulation will not be a privilege for big tech (e.g., Tesla). Let academic democratize humanoids. web: humanoid-teleop.github.io

Introducing LocoTouch: Quadrupedal robots equipped with tactile sensing can now transport unsecured objects — no mounts, no straps. The tactile policy transfers zero-shot from sim to real. Core Task-Agnostic Features: 1. High-fidelity contact simulation for distributed tactile