Songlin Wei (@SonglWei) profile picture

PhD Student in Robotics & Embodied AI @USC |prev. @pku1898

Los Angeles, CA
Joined June 2020
Photo by Songlin Wei (@SonglWei)Photo by Songlin Wei (@SonglWei)
Truly exciting to see humanoids advancing into the loco-manipulation era!
Introducing Ψ₀ (psi-lab.ai/Psi0) — an open foundation model for universal humanoid loco-manipulation. 🏆 Outperforms GR00T N1.6 by 40%+ overall success rate 📉 Uses only ~10% of the pre-training data 📦 Fully open-source: model, data, code, and deployment pipeline 1/10
3
5
449
Songlin Wei retweeted
The @Tesla Roadster stunts, simulated in MuJoCo! Thrusters are site actuators, the ramp suction uses adhesion actuators, and the builtin box QP decides how hard to fire each jet.
13
25
6
361
19,880
🔥 SIMPLE is accepted to CoRL 2026! 🤖 Tired of the complexity and cost of setting up whole-body teleoperation on real robots? With SIMPLE, you can bring SONIC whole-body teleoperation directly into simulation, enabling up to 3× higher data collection efficiency. ⚡ What SIMPLE offers: • Whole-body humanoid teleoperation with SONIC • Up to 3× faster data collection • Fast, parallel policy rollouts in simulation • A unified benchmark for mainstream VLA / World-Action Models, including GR00T, DreamZero, Cosmos3, π₀, and π₀.₅ Collect faster. Evaluate faster. Iterate faster — all in simulation. Check it out! 👇 📎 Project Page: psi-lab.ai/SIMPLE/ 📄 Paper: arxiv.org/abs/2606.08278 💻 Code: github.com/physical-superint…
4
12
1
114
7,897
Songlin Wei retweeted
🔥Diffusion Masters Transparent Objects in Videos! We introduce #DKT, a foundation model that repurposes video diffusion for zero-shot depth and normal estimation on Transparent and Reflective Objects with Superior Temporal Consistency Demo: huggingface.co/spaces/Daniel… Code: github.com/Daniellli/DKT
6
40
5
302
28,211
Songlin Wei retweeted
"Who has seen the wind? Neither I nor you: But when the leaves hang trembling, the wind is passing through." @ScottZhiyuanGao will present our work "Seeing the Wind from a Falling Leaf" at #NeurIPS2025! We teach AI to "see" invisible force fields directly from video using differentiable physics. Project Page: chaoren2357.github.io/seeing… Paper: arxiv.org/abs/2512.00762
5
5
3
33
12,013
Are SoTA monocular methods ready for robotic tasks yet? Check this demo out! For precise robotic manipulation tasks, our stereo-depth method still outperforms MDE in OOD scenarios:
Introducing D3RoMa: a diffusion model-based stereo depth estimation method capable of restoring missing depths for transparent and reflective objects. Code release at pku-epic.github.io/D3RoMa/! Work collaborated with @HaoranGeng2 @CongyueD.
1
20
1
123
15,734
@HaoranGeng2 @CongyueD Check this out, new demo :)
1
465
Introducing D3RoMa: a diffusion model-based stereo depth estimation method capable of restoring missing depths for transparent and reflective objects. Code release at pku-epic.github.io/D3RoMa/! Work collaborated with @HaoranGeng2 @CongyueD.
3
14
4
65
22,746
Countdowns to top CV/NLP/ML/Robotics/AI conference deadlines aideadlin.es/?sub=ML,CV,CG,N…
2