Haocheng Yin

Ph.D. Student @GeorgiaTech

[Haocheng Yin @hcy1n]

@notprajyoth The material estimation module was trained under a synthetic dataset augmented from a baseline dataset, where the material properties are randomized for each synthetic demo. It was then tested on unseen real-world demos (shown in the video) and the dynamics correction works well.

[Haocheng Yin @hcy1n]

Excited to share our #CoRL2026 work PhysCoRe: a physics-corrected world model for deformable objects!

We couple a physics-based simulator with two feed-forward modules: one infers material from vision, the other corrects its internal dynamics.

Check: https://t.co/NWLKcYaUc8

[Haocheng Yin @hcy1n]

@aneeshers Thank you Aneesh!

[Haocheng Yin @hcy1n]

Huge thanks to my collaborator @ShuohanT, whose leading contribution made this work possible!

[Haocheng Yin @hcy1n]

Thrilled to share our latest research work on generative predictive control (GPC)! It’s been an incredible journey working on this project with @hanqi359246, Prof. Yilun @du_yilun and Prof. Hank @hankyang94 at Harvard SEAS! Really grateful for their supervision and collaboration!

[Heng Yang @hankyang94]

Diffusion has shown great promise for generating robot **actions**, can it act as a **world model** to generate the future conditioned on actions?

In our work led by @hanqi359246 @hcy1n and in collaboration with @du_yilun, we show a **controllable** action-conditioned video diffusion model can produce photorealistic and (near) physics-accurate future predictions.

This ability strengthens the policy via:

- ranking different action proposals and selecting the best, or

- **visual** trajectory optimization by optimizing the action proposals using gradient ascent.

Learn more about Generative Predictive Control (GPC) at:
https://t.co/0BeujMLhVW

[Heng Yang @hankyang94]

Robots in our group are finally moving!
Excited to share the first #RobotLearning project from my group led by Han Qi and @hcy1n:

Control-oriented Clustering of Visual Latent Representation

In classical robotics, "perception" and "control" are separated with an explicit state representation.

In end-to-end vision-based control, what's going on in the visual latent representation space?

- Focusing on the planar pushing task, we found an elegant geometric structure in the visual latent space: visual features are clustered according to "control-oriented" goals. Moreover, such clustering has a rigid structure known as "neural collapse" (NC, first found in the image classification literature).

- How does this understanding help? We show that pretraining the vision encoder with control-oriented clustering significantly increases test-time performance in the low-data regime (e.g., 50 or 100 demos). For example, with 100 demos, baseline succeeds 5 out of 10, but NC-pretrained policy succeeds 8 out of 10.

Check out more about the project:
https://t.co/b9UeCSoPmf