YC S25 A hub for showcasing, testing, and advancing robotics technology in the heart of Silicon Valley. Pre-order M1 now: https://nitter.cf/t.co/QSIWFy2CzV
San Francisco, CA
Joined October 2025
- Tweets149
- Following19
- Followers528
- Likes33
LLM vs. WAM: Astra vs. Dream-1. Round 3.
Cup stacking. Same arms. Same task. Zero-shot.
A: 1:06
B: 0:41
Both finish. Which one is the LLM?
A or B? Defend your pick.
#Robotics #EmbodiedAI
RAX Learns from Failure: 136s → 96s.
Real-robot exploration turns failed attempts into executable skills.
No model calls during execution.
RAX × Robotics Center
#RAX #Robotics #EmbodiedAI
Robot Hackathon with @MongoDB and @viamrobotics at Robotics Center — Day 1.
Builders are coding, testing, and debugging on real robot arms.
The room is busy building. Demos tomorrow.
#Robotics #PhysicalAI #Hackathon
Your code. Real robot arms. 48 hours to build.
Robot Hackathon with @MongoDB and @viamrobotics at Robotics Center, Oct 29–30.
Build physical AI applications using real hardware, Viam and MongoDB Atlas.
Event details: luma.com/ta40197s
#PhysicalAI #Robotics
LLM vs. WAM: GPT-6 Astra vs. Dream-1
Which one would you bet on?
A or B — which one is the LLM?
Reply with your guess and tell us why.
Correct guesses enter a mystery gift drawing.
@sama @jang_yoel @GeYunhao @scott_e_reed @youliangtan
#Robotics #EmbodiedAI #LLM
LLM vs. WAM: Astra vs. Dream-1. Round 2.
Same bimanual arms. Same marker handover task. Zero-shot.
A fails. B succeeds.
Which one is the LLM?
A or B? Defend your pick.
#Robotics #EmbodiedAI
An agent explores on a real robot for 2 hours, then runs the skills it learned as code.
That’s RAX.
No demonstrations. No teleoperation. Just exploration turned into reusable robot skills.
@RunhanH @xczhangxx @du_yilun @pabbeel @wangyenjen
#Robotics #EmbodiedAI
Can an LLM control a real robot zero-shot?
GPT-6 Astra × xArm × Wuji Hand 2.
Locate. Plan. Grasp. Lift. Release.
Cube into bowl. Bottle upright.
No task-specific demos.
How far can this approach scale?
#Robotics #EmbodiedAI
The 2026 Robotics Center Research Awards call for proposals is open through December. Awards provide hardware credits and robot time for robotics and robot learning research that depends on physical robots. Rolling review.
eval.roboticscenter.ai/resea…
Agents like RAX (@du_yilun @RunhanH @pabbeel @wangyenjen @Lyubh22) now teach robots by trial and error.
The brain isn't the bottleneck anymore. The loop is.
We build Physical RSI infrastructure. 100+ robots, one referee: eval.roboticscenter.ai
We build Physical RSI infrastructure. 100+ robots, one referee: eval.roboticscenter.ai
Same AI brain. 3 different robot bodies. Does it still work?
We ran 4 frontier LLMs on an AgileX Piper, a UFactory xArm and an I2RT YAM. Same task, same prompt, same action space.
The body changes how the brain fails
🤖 What if robot evaluation worked like DoorDash?
🧠 Pick an AI model
🦾 Choose a robot
🎯 Set a task
📊 Get real-world results
We're building Evaluation-as-a-Service at Robotics Center.
🔗 eval.roboticscenter.ai/
#Robotics #PhysicalAI
Can robots improve without retraining?
Join Saturday Robotics × AIM Intelligence tomorrow in SF for Robotics & World Models Reading Club #34! Explore RPG, robot learning & embodied AI with Bay Area researchers and builders.
📅 Oct 17
🔗 luma.com/oq82q5ax
#Robotics #AI
Excited to host @viamrobotics @MongoDB robotics hackathon at our space in SF!
Every member of the winning teams will take home their own SO-101 robot kit, plus Atlas credits (up to $5,000 of each).
📅 Oct 29-30 📍 Robotics Center of Silicon Valley, San Francisco
Apply now: luma.com/ta40197s?utm_campai…
Robotics Center of Silicon Valley retweeted
Went to Dexterity Day at @RoboticsCtr yesterday.
Yes, on Saturdays we also do tech.
Around 200 people talking robot hands, data and deployment, and great meeting new people like @Lakshya__Bakshi and @abhihjoshi 🧵
