Simon LC

research scientist @ the AI institute
robotics & optimization phd @ stanford

[Simon LC @simonlc_]

Excited to share what I've been working over the last few months! We taught Spot to lift, roll, and stack tires completely autonomously. It uses its arms, body, and legs to manipulate heavy objects at speed.

[Simon LC @simonlc_]

If you're interested in working on whole-body manipulation at @rai_inst feel free to use this link https://t.co/qCEVYkH5sr
Check out the blog post for more details: https://t.co/9vPJXQmyYT

[Simon LC @simonlc_]

Excited to share what I've been working over the last few months! We taught Spot to lift, roll, and stack tires completely autonomously. It uses its arms, body, and legs to manipulate heavy objects at speed.

[Simon LC @simonlc_]

We also show RL-based tire uprighting, where the tire literally flies into the air. This illustrates just how dynamic whole-body manipulation can be.

[Simon LC @simonlc_]

I am really happy to finally share this work! Huge congrats to @albert_h_li and @pdculbert for the effort they put in making the tool hackable and interactive.
Try it!🐍
pip install judo-rai

[Preston Culbertson @pdculbert]

🥋 We're excited to share judo: a hackable toolbox for sampling-based MPC (SMPC), data collection, and more, designed to make it easier to experiment with high-performance control.
Try it: pip install judo-rai

[Kuan Fang @KuanFang]

Our new paper shows how task representations learned via temporal alignment enable compositional generalization for conditional policies. This allows robots to solve compound tasks by implicitly decomposing them into subtasks.

[Xuanlin Li (Simon) @XuanlinLi2]

Learning bimanual, contact-rich robot manipulation policies that generalize over diverse objects has long been a challenge.

Excited to share our work: Planning-Guided Diffusion Policy Learning for Generalizable Contact-Rich Bimanual Manipulation!

https://t.co/df7t8bJs2m
🧵1/n

[Preston Culbertson @pdculbert]

ICYMI: For #CoRL2024 we released a dataset of 3.5M (!) dexterous grasps, with multi-trial labels and perceptual data for 4.3k objects.

Our takeaways: scale matters, and refining grasps > better sampling. Hoping our data can enable more vision-based grasps in hardware!

[Simon LC @simonlc_]

We're presenting Jacta: a versatile planner for learning dexterous and whole-body manipulation this week at CoRL!
website https://t.co/iHZpdakP0P
paper https://t.co/xhYWl9BTvg

[RAI Institute @rai_inst]

Our team is presenting work at the Conference on Robot Learning, @corl_conf, in Munich, Germany this week! Learn more about our accepted research — https://t.co/438na24z7y

[Simon LC @simonlc_]

This is a joint work with an amazing team! Jan Brüdigam, Ali Abbas, @initmaks, @KuanFang, Brandon Hung, Maya Guru, Stefan Sosnowski, Jiuguang Wang and Sandra Hirche
Jan did an awesome job leading the project during his internship at @the_ai_inst. So proud of what we accomplished!

[Simon LC @simonlc_]

I'm excited to share Jacta: A Versatile Planner for Learning Dexterous and Whole-body Manipulation.
We use sampling-based planning to bootstrap policy learning methods for manipulation tasks.
My friend, Jan Brüdigam is presenting the work today at CoRL!
https://t.co/iHZpdakP0P

[Simon LC @simonlc_]

We used reinforcement learning bootstrapped with expert planner demonstrations to learn robust policies. We deployed them on several hardware scenarios using Boston Dynamics' Spot

[Albert Li @albert_h_li]

Excited to share our new📰, DROP: Dexterous Reorientation via Online Planning!

Overview:
🔹We tackle cube rotation🧊♻️on hardware
🔹DROP is the first 🧊♻️sampling-based MPC demo. No reinforcement learning!
🔹Median 30.5 rotations w/o dropping, max of 81👑🦾

See 🧵below👇

[Simon LC @simonlc_]

Really cool work on bimanual dexterity!!

[Toru @ToruO_O]

Achieving bimanual dexterity with RL + Sim2Real!

https://t.co/AIckktMuTq

TLDR - We train two robot hands to twist bottle lids using deep RL followed by sim-to-real. A single policy trained with simple simulated bottles can generalize to drastically different real-world objects.

[Naoki Yokoyama @naokiyokoyama0]

Excited to share our latest work, Vision-Language Frontier Maps – a SOTA approach for semantic navigation in robotics. VLFM enables robots to navigate and find objects in novel environments using vision-language foundation models, zero-shot! Accepted to #ICRA2024!
🧵

[John Zhang @johnzhangx]

want your next robot 🤖 to move like your favorite pet 🐶? youtube videos might be all you need.

checkout our recent work -- SLoMo

paper: https://t.co/uKYne7ZenU
website: https://t.co/O2k6mezd4N
video: https://t.co/uzc85nW3sk