bikes on saturday, hardware @ sunday
San Mateo
- Tweets 1,773
- Following 1,317
- Followers 5,523
- Likes 11,572
if you're wondering how dire robotics manufacturing in the US is:
in the last 2 months there have been 3 robotics startups launch that just wrap a Unitree G1
22
21
192
9,405
co-optimizing hardware, controls, and ML is extremely powerful in robotics
Transformer Transformer can condition on diverse motions to co-optimize robots for them. Here, put in an UMI dataset, get out an optimized mobile manipulator! Turns out @sundayrobotics's Memo torso design was critical 🤔 @chengchicc proud of you hehe and thank you ❤️🔥🙏
2
5
129
15,726
clayton retweeted
In the past 6 months we’ve produced a cheaper, more efficient seeding material, trained a hyper-resolution AI weather model, and operated 100+ cloud seeding flights for the governments of Texas and New Mexico.
Recast Systems is coming out of stealth. Here’s some of the research we’ve been working on. (1/n)
84
108
1,014
335,793
incredibly smart to have a concave structure with guy wires to load it in compression rather than bending
Mtech Operations, an MIT team using autonomous flight, has now moved into 2nd place with a 3.41:1 ratio
One of the most impressive showings, and it's not gas or a helicopter style
32.8 lb drone
111.9 lb payload (3.41:1)
Completed the course in under 24 minutes
6
12
553
66,193
1000 strong! Was such a pleasure to work closely with the first bunch, thanks for sticking with us.
1/N Two years ago, we recruited our first Memory Developer off Craigslist and onboarded them in a public library.
Today, more than 1,000 Memory Developers have helped us build the data engine behind ACT-2.
Here's how we got from that library to here 🧵
3
53
3,374
clayton retweeted
I came from a world where I measured robotics research in days: collect hundreds of demonstrations, train, then watch the policy break as soon as the setup changed.
Since I joined Sunday, I’ve watched that loop compress dramatically.
ACT-2 can learn a new folding strategy from one demonstration, with just five minutes of fine-tuning, then transfer it to unseen garments and beds. What once took a week can now happen in an hour.
That’s the deeper impact of scaling pretraining: it doesn’t just produce a better policy. It produces a better learner.
Introducing ACT-2 Preview
The first robotics model to unify broad generalization with high reliability. A single fine-tuning example can teach Memo a new behavior that generalizes.
Zero shot, real unseen homes, 99% success rate.
14
10
156
17,442
Its been extremely exciting to come into the office everyday and watch ACT-2 do progressively more and more!
At this point, I don't even ask if its in the dataset anymore - the answer is almost always no.
Introducing ACT-2 Preview, the world’s first robotics model that works in your home.
99% success rate, fully autonomous in unseen homes. Zero data from you.
1
3
84
3,784