Rhythm Syed
@rhythm_syed
Followers
7
Following
28
Media
1
Statuses
9
CS PhD Student @Columbia & Researcher @ToyotaResearch #Robotics #AI
Palo Alto, CA
Joined January 2025
[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: 🎮 fine-grained teleop for novices 🤖 a copilot learned from <5 min of
1
2
6
Grateful to have had the opportunity to present Interactive World Simulator and some teasers for my upcoming VLA work with @ToyotaResearch at the Computing Research Association Grad Cohort Workshop! Thank you to @YunzhuLiYZ for being an amazing advisor and guiding my research!
0
0
2
World models are becoming central in robotics. But most still struggle with two things: speed and long-horizon physical consistency. A new system called Interactive World Simulator pushes both forward. It’s an action-conditioned world model that can generate 10+ minutes of
0
11
62
Even my parents with NO technical background can play with our web demo in China!! It might be one of the proudest moments of my PhD to build something that my parents can access and understand. That's also exactly one of our motivation: build WMs accessible for everyone!
You can actually interact with the world simulator directly in the browser. 🤖 Here is a quick screen recording (8x speed) of me playing with it: real-time action-conditioned video prediction across rigid objects, deformable objects, rope, and object piles. Try it yourself (no
5
6
141
You can actually interact with the world simulator directly in the browser. 🤖 Here is a quick screen recording (8x speed) of me playing with it: real-time action-conditioned video prediction across rigid objects, deformable objects, rope, and object piles. Try it yourself (no
For a long time, I was skeptical about action-conditioned video prediction for robotics. Many models look impressive, but once you ask them to handle long-horizon manipulation with real physical interaction, things quickly fall apart (e.g., Genie is amazing but mostly focused on
1
29
234
Wow, love the results here! Also a really exciting direction for general reward modeling. The fact that things can be plug-and-play here could unlock a lot of potential for using our Interactive World Simulator for RL and policy learning.
Really cool interactive simulator! Love that you can also added a reward model to it , excited to see what we can build with this.
1
6
36
For a long time, I was skeptical about action-conditioned video prediction for robotics. Many models look impressive, but once you ask them to handle long-horizon manipulation with real physical interaction, things quickly fall apart (e.g., Genie is amazing but mostly focused on
1/ World models are getting popular in robotics 🤖✨ But there’s a big problem: most are slow and break physical consistency over long horizons. 2/ Today we’re releasing Interactive World Simulator: An action-conditioned world model that supports stable long-horizon interaction.
2
51
363
1/ World models are getting popular in robotics 🤖✨ But there’s a big problem: most are slow and break physical consistency over long horizons. 2/ Today we’re releasing Interactive World Simulator: An action-conditioned world model that supports stable long-horizon interaction.
24
83
472