We might be solving the wrong problem in robotics.
That’s what this makes clear.
UMI → Universal Manipulation Interface
A simple $400 gripper that lets you teach robots by demonstration.
You hold it like a tool. Show the task. The robot learns.
No teleoperation.
No expensive hardware.
No robot-specific data.
Stanford open-sourced everything → hardware, code, datasets.
What stands out to me is the bottleneck.
Not algorithms.
Data.
Teleoperation → ~35 demos/hour
UMI → ~111 demos/hour
And the data transfers across robots → UR5, Franka, others.
The design is surprisingly practical:
→ GoPro fisheye lens (155° FOV) + mirrors for depth
→ SLAM + IMU for precise 6DoF tracking
→ latency matching for dynamic tasks
→ diffusion policies for multimodal actions
Then it scales.
Cheng Chi takes this further with Sunday Robotics (with Tony Zhao).
A $200 glove → deployed in 500+ homes → ~10 million real-world interactions.
Not lab data.
Real human behavior.
Their robot learns dishes, laundry, espresso → with zero robot-specific data.
This is where the shift becomes obvious.
From training robots in controlled environments
→ to learning directly from humans at scale
So here’s the real question:
Will robotics be unlocked by better models… or by unlocking data?
#ArtificialIntelligence #Robotics #AI #Innovation #FutureOfWork
@XMihura Pinta bastante bien! Yo conecto claudecode a elevenlabs para recibir notificaciones del progreso de los agentes o su finalizacion. Ayuda a distraerse menos ;)
i have no desire to be rich so i can buy a lambo or birkins.
I want to be rich so I can control my time and go to the gym at 2 pm on a wednesday.
sit at a cafe and relax for an hour on a rainy afternoon.
so I can cook meals at home with fresh ingredients.
spend on my family and friends without worrying about a budget.
that’s my idea of a rich life, not the fake consumerist idea shoved down my throat.
Como es el pipeline que te hiciste en Obsidian? Yo por ahora voy guardando de una manera estructurada todo lo que voy generando con claudecode, pero a veces me cuesta saber como aprovecharlo después más que para tenerlo como contexto en la proxima sesion. Me interesa mucho saber como lo haces! :)
I keep thinking about this. I used claude code to analyze my dna data + blood&urine analysis i did recently + what I eat + what I do when I workout. The results have been pretty amazing. I have a health protocol including habit recommendations like what to do when I wake up, doing NSDR, when to workout, when I am able to focus more, etc. Combined with a supplementation (magnesium, omega-3, creatine, etc.) action plan and some additional recommended tests. I think people could really benefit from a tool that did this and they would have a clear idea of what THEY need instead of trying supplements just because Huberman said to do so.
Bastante, pero bastante impresionante, el espectáculo por el año nuevo chino con robots de Unitree haciendo acrobacias en combo con niños!
Es impresionante lo que han evolucionado en sólo un año. Disfrutadlo, porque estamos a esto 🤏 de normalizarlo.
15/ If there's enough interest, I'll package this up as a service.
Setup + custom configuration + integrations (WhatsApp, reminders, etc).
Would you use something like this? Reply or DM me - genuinely curious if others have this "disorganized brain" problem.
I've always wanted to take notes, but I'm too disorganized.
I'd write things down, then lose track. Notes scattered everywhere.
So I built something different: @claudeai as my personal assistant that structures my entire life automatically in @obsdmd
Here's the setup 🧵
14/ The future of productivity isn't trying harder.
It's building better systems.
Huge credit to @kepano for sharing his work publicly and making this possible. Standing on the shoulders of giants.
The "where did I write that?" era is over.