James (Jingxi) Xu

senior research scientist @antgroup and postdoc @stanford. ex @rai_inst @columbia @mit @penn. robot learning.

[Litian Liang @litian_liang]

UME is accepted to CoRL 2026 🎉
https://t.co/W3PHmYRm4q

[James (Jingxi) Xu @jamesjingxi]

@thejonohart @EdmondIsARobot @angelajiazhang @litian_liang @binghao_huang @kaiwynd @bardienus @Darpinian @adamraudonis @PaulKefer @Kylabearrrr We will miss you!

[Edmond @EdmondIsARobot]

A huge thank you to everyone who joined
@SundaeRobotics 03! 🤖🍨 Special thanks to @binghao_huang for an outstanding talk on the future of flexible tactile skin for dexterous manipulation.

🤖🍨 Next Sundae (https://t.co/WqKLdxQag1), we're excited to host @kaiwynd, a Ph.D. student at Columbia University advised by Prof. Yunzhu Li, for a talk on engineering robotic simulators for evaluation, data generation, and sim-to-real robot learning.

Thanks to @angelajiazhang @thejonohart @menemazarakis @drjingxi @litian_liang @Joshuabrowder

[James (Jingxi) Xu @jamesjingxi]

RT @EdmondIsARobot: A huge thank you to everyone who joined Sundae Robotics 02. 🤖🍨

Special thanks to @ChanghaoWang78 for an outstanding ta…

[Tanay Tandon @tanaytndn]

one step closer to becoming a clanker

[James (Jingxi) Xu @jamesjingxi]

Thanks for hosting!!

[angela zhang @angelajiazhang]

great first Sundae Robotics 🙃 thanks @drjingxi for speaking on his work with UME - we’ll be doing this every Sunday and also starting a Boston series. rsvp to next sunday’s in comments; reach out if you’re working on robotics and interested!

@EdmondIsARobot @menemazarakis

[James @Darpinian]

Universal Manipulation Exoskeleton presentation at Sundae Robotics reading group. They brought the katana...

[James (Jingxi) Xu @jamesjingxi]

Great capture!

[James @Darpinian]

Robot gets a beer from the fridge

[James (Jingxi) Xu @jamesjingxi]

Congrats! Very cool work!

[Kelin Yu @ColinYu14116982]

OmniTacTune won the Best Paper Award at the Tactile Sensing for Robotic Foundation Models Workshop at RSS 2026! (https://t.co/N3pNabWg5K)

Huge thanks to the coauthors @HaodeZ47056 and @HanYunhai, and our advisors @h_ravichandar and @RuohanGao1.

[James (Jingxi) Xu @jamesjingxi]

Looking forward to talking about UME at this event. We will also bring the hardware so people can check it out in person!

[Edmond @EdmondIsARobot]

🤖🍨 Sharing the first Sundae Robotics event, a new series bringing together a tight-knit group of robotics researchers, founders, and builders for frontier talks, technical discussions, and sundaes.

The inaugural session features one of the main authors behind the recent Universal Manipulation Exoskeleton (UME) paper.

[James (Jingxi) Xu @jamesjingxi]

@YXWangBot @HaonanChen_ @YunzhuLiYZ @Zhou_Yu_AI @dineshjayaraman @Majumdar_Ani Congrats Dr. Wang!!

[James (Jingxi) Xu @jamesjingxi]

@JunyaoShi @sundayrobotics Congratsss !!

[James (Jingxi) Xu @jamesjingxi]

@simon_zhai Thank you, Simon!

[James (Jingxi) Xu @jamesjingxi]

Can't agree more!

[Litian Liang @litian_liang]

"Computer Vision is a Robotics problem, but Robotics is not a Computer Vision problem." - I learned this from Professor @SongShuran's lecture in 2024 at @Stanford.

The box pushing task in our recent work UME is yet another demonstration of this point: https://t.co/eQOqNCgQzV uses torque feedback to give the robot model direct input to determine if a box is pushed to the end of a constrained, visually occluded environment.

This task is designed by @drjingxi to kill vision only robot policies, and he's done it.

In fact, I frequently bring up how I think this task is so brilliantly designed and my favorite task among the 4 autonomous model experiment we did. So I might as well do it here on https://t.co/UpfxMK3uun.

A great experiment is not one that shows existing method can not solve a hard problem, but can not solve a simple one, one that we did not realize it can't do.

Looks so simple yet so hard. That's why it's so great.

[James (Jingxi) Xu @jamesjingxi]

Thanks @IlirAliu_ for sharing our work! Love your analysis!

[Ilir Aliu @IlirAliu_]

This low-cost exoskeleton could make robot training 10x faster:

A low-cost exoskeleton that provides real-time haptic torque feedback, allowing a human teleoperator to feel and control a robot’s forces during complex manipulation tasks while capturing data for autonomous policies.

Demonstrations include a blindfolded user unsheathing a sword via the mirrored robot arm and autonomous robots performing force-sensitive tasks like retrieving a drink from a fridge side holder or navigating tight spaces in a kitchen.

The system uses bidirectional kinematics and dynamics retargeting for compatibility with different robots.

Enabling rapid data collection, overnight training, and next-day autonomy for whole-body, occluded, and long-horizon mobile manipulation.

Thanks for sharing this great work, @litian_liang!

📌 https://t.co/S3XJKXI3NF​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​

——

Weekly robotics and AI insights.
Subscribe free: https://t.co/9Nm01QUcw3

[James (Jingxi) Xu @jamesjingxi]

Hi UMe tea, DM me for our bank account numbers. Venmo works too.

[Litian Liang @litian_liang]

Achievement Unlocked:

Drinking UME while wearing UME

https://t.co/5mJr6E0L4w

[James (Jingxi) Xu @jamesjingxi]

Thanks Sholto for sharing our work and your thoughts! I’m very much on board with what you’re saying. Also really looking forward to what @AnthropicAI’s newly formed robotics team will do.

[Sholto Douglas @_sholtodouglas]

As the humanoid supply chain develops it’ll unlock ready player one style force feedback exoskeletons - big opportunity to build something cool here in the next few years.
Of course - likely humanoids will outbid VR for every marginal motor until 2030+, but see you in VR Skyrim then 😎

[James (Jingxi) Xu @jamesjingxi]

@huaijiangzhu Thanks, Huaijiang! Come and try it in person next time you are in the bay.

[James (Jingxi) Xu @jamesjingxi]

Thanks, Yixuan! Things have been moving very fast, so there may have already been some iterations since your visit. You’re always very welcome to come by again and try it in person!

[Yixuan Wang @ IROS 2026 @YXWangBot]

Congrats on the release!! I have seen this setup in person and it is very awesome! Cannot wait to try it!

[James (Jingxi) Xu @jamesjingxi]

Huge thanks @chris_j_paxton for sharing our work and your thoughts 🙏🙏 Always impressed by how deeply you understand robotics!

[Chris Paxton @chris_j_paxton]

The kind of things you can do with compliant/force aware teleop i guess. Another cool video of a highly constrained pick.