Shuo Sha Profile
Shuo Sha

@shashuo0104

Followers
48
Following
14
Media
5
Statuses
19

Undergrad at @Columbia #Robotics

New York, USA
Joined July 2024
Don't wanna be here? Send us removal request.
@shashuo0104
Shuo Sha
4 hours
[1/5] Fine-grained teleop is slow, error-prone, and frustrating even for experts. We introduce a real2sim2real shared autonomy framework that learns a residual copilot for low-level corrections. It enables: ๐ŸŽฎ fine-grained teleop for novices ๐Ÿค– a copilot learned from <5 min of
2
6
36
@shashuo0104
Shuo Sha
2 hours
Thanks Kaifeng for sharing!! As an embodiment-agnostic framework, Iโ€™m also curious to see how Residual Copilot might apply to other robotic platforms.
@kaiwynd
Kaifeng Zhang
3 hours
Despite the success of whole body RL for humanoid teleoperation, teleoperation for manipulation (parallel jaw gripper) has mostly just been direct joint space or Cartesian space mapping. This new work from our lab shows that for contact-rich manipulation such as industrial
0
0
1
@YunzhuLiYZ
Yunzhu Li
4 hours
Teleoperation is often used to scale robot data collection. But anyone who has actually done fine-grained teleop knows how hard it is, even for experienced operators: precise alignment, axis-constrained rotation, contact regulation, and millimeter-level control are extremely
@shashuo0104
Shuo Sha
4 hours
[1/5] Fine-grained teleop is slow, error-prone, and frustrating even for experts. We introduce a real2sim2real shared autonomy framework that learns a residual copilot for low-level corrections. It enables: ๐ŸŽฎ fine-grained teleop for novices ๐Ÿค– a copilot learned from <5 min of
0
3
29
@shashuo0104
Shuo Sha
4 hours
Huge thanks to all my collaborators @YXWangBot @binghao_huang for making this work possible!! Grateful to my advisors @antoniloq @YunzhuLiYZ for their guidance!!
0
1
5
@shashuo0104
Shuo Sha
4 hours
Links โ†“ Website: https://t.co/q6eeJ0s8Dv Sim code: https://t.co/vsyMlalOCH Deploy code: https://t.co/lmfHRQrtvO Paper:
1
1
4
@shashuo0104
Shuo Sha
4 hours
[5/5] Challenge #2: Assistance must respect user control authority. A residual copilot makes only local corrections -> preserving user intent. We visualize the learned residual behaviors in simulation.
1
1
3
@shashuo0104
Shuo Sha
4 hours
[4/5] Challenge #1: Learning assistive behaviors in simulation requires a human surrogate that: โ€ข uses very little data โ€ข behaves human-like in unseen states Our solution: a simple yet surprisingly effective kNN human surrogate.
1
1
4
@shashuo0104
Shuo Sha
4 hours
[3/5] Copilot-assisted teleop produces better demonstrations. With the same number of demos, imitation policies trained on copilot data perform significantly better. The copilot improves the structure and consistency of successful trajectories.
1
1
4
@shashuo0104
Shuo Sha
4 hours
[2/5] The residual copilot improves performance across assembly tasks: ๐Ÿ”ฉ Nut Threading: 40% -> 100% success โš™๏ธ Gear Meshing: 16.4s -> 10.9s completion ๐Ÿ“Œ Peg Insertion: 30.3s -> 18.5s completion Observed across different subjects in our user study.
1
1
4
@shashuo0104
Shuo Sha
5 days
Pixel-based world models are clearly the most scalable path, and @YXWangBot is proving that we donโ€™t have to sacrifice physical consistency to get there! The visual appearances are so convincing that I keep getting fooled on which ones are real vs generated.
@YXWangBot
Yixuan Wang
6 days
1/ World models are getting popular in robotics ๐Ÿค–โœจ But thereโ€™s a big problem: most are slow and break physical consistency over long horizons. 2/ Today weโ€™re releasing Interactive World Simulator: An action-conditioned world model that supports stable long-horizon interaction.
0
0
7
@stepjamUK
Stephen James
4 months
๐—ง๐—ฒ๐˜€๐˜๐—ถ๐—ป๐—ด ๐—ฟ๐—ผ๐—ฏ๐—ผ๐˜ ๐—ฝ๐—ผ๐—น๐—ถ๐—ฐ๐—ถ๐—ฒ๐˜€ ๐—ถ๐—ป ๐˜๐—ต๐—ฒ ๐—ฟ๐—ฒ๐—ฎ๐—น ๐˜„๐—ผ๐—ฟ๐—น๐—ฑ ๐—ถ๐˜€ ๐—ฒ๐˜…๐—ฝ๐—ฒ๐—ป๐˜€๐—ถ๐˜ƒ๐—ฒ, ๐˜€๐—น๐—ผ๐˜„, ๐—ฎ๐—ป๐—ฑ ๐—ต๐—ฎ๐—ฟ๐—ฑ ๐˜๐—ผ ๐—ฟ๐—ฒ๐—ฝ๐—ฟ๐—ผ๐—ฑ๐˜‚๐—ฐ๐—ฒ. ๐—•๐˜‚๐˜ @Columbia ๐—ฎ๐—ป๐—ฑ @SceniXai ๐—ท๐˜‚๐˜€๐˜ ๐—ฏ๐˜‚๐—ถ๐—น๐˜ ๐—ฎ ๐˜€๐—ถ๐—บ๐˜‚๐—น๐—ฎ๐˜๐—ผ๐—ฟ ๐˜๐—ต๐—ฎ๐˜ ๐—ฎ๐—ฐ๐˜๐˜‚๐—ฎ๐—น๐—น๐˜† ๐˜„๐—ผ๐—ฟ๐—ธ๐˜€. They
4
8
54
@kaiwynd
Kaifeng Zhang
4 months
If you are working on real-to-sim, simulating digital twins, and policy evaluation, you should check out our fully open-sourced code base. Lots of handy tools for building Gaussian Splatting simulators and interacting with it! https://t.co/Ovpm0hFdYn Will continue to be
Tweet card summary image
github.com
Open-source code of the paper: Real-to-Sim Robot Policy Evaluation with Gaussian Splatting Simulation of Soft-Body Interactions. - kywind/real2sim-eval
4
40
289
@shashuo0104
Shuo Sha
4 months
Thanks @kaiwynd! Super excited to share our unified policy training & inference repo -- used to train Diffusion Policy, ACT, Pi0, and SmolVLA, and to rollout policies in both the real-world and our real2sim simulator (available at https://t.co/Kmya0tWIxZ) policy training repo:
@kaiwynd
Kaifeng Zhang
4 months
Also check out our policy training repo: supporting two main frmeworks, Lerobot @LeRobotHF and OpenPI @physical_int , with a unified data & inference interface. https://t.co/zht92pZ47Z Made by my amazing collaborator @shashuo0104
0
0
2
@YunzhuLiYZ
Yunzhu Li
4 months
I want to call out one of our most important references: SIMPLER ( https://t.co/8xoHzaF46Z), by @XuanlinLi2, @kylehkhsu, @Jiayuan_Gu, @jiajunwu_cs, @haosu_twitr, @QuanVng, @xiao_ted, and colleagues, which laid the foundation for using simulation for policy evaluation through a
@YunzhuLiYZ
Yunzhu Li
4 months
๐Ÿ“ข Announcing one of the most exciting works from us this year on **scalable robot policy evaluation through real-to-sim transfer**, moving toward a scalable evaluation engine with structured world models that capture the appearance, geometry, and dynamics of environments
5
11
95
@YunzhuLiYZ
Yunzhu Li
4 months
๐Ÿ“ข Announcing one of the most exciting works from us this year on **scalable robot policy evaluation through real-to-sim transfer**, moving toward a scalable evaluation engine with structured world models that capture the appearance, geometry, and dynamics of environments
@kaiwynd
Kaifeng Zhang
4 months
๐Ÿงต Evaluating robot policies in the real world is slow, expensive, and hard to scale. During my internship at @SceniXai this summer, we had many discussions around the two key questions: how accurate must a simulator be for evaluation to be meaningful, and how do we get there?
2
30
213
@kaiwynd
Kaifeng Zhang
4 months
๐Ÿงต Evaluating robot policies in the real world is slow, expensive, and hard to scale. During my internship at @SceniXai this summer, we had many discussions around the two key questions: how accurate must a simulator be for evaluation to be meaningful, and how do we get there?
8
32
151