量産系の工場に身を置くひとりとして、こういう取り組みは面白いなと思う。
ただ、何でもかんでもAIにすればいいって話でもないんですよね…。
今回のような飲料補充タスクなら、再現性や安全性、保守性、コストを考えると、従来の画像処理やロボット制御で十分対応できる部分も多そうやし。
コンビニ大手のファミリーマートは、飲料を自動で補充するフィジカルAIを活用したロボットを本格導入する方針です。どんなロボットなのでしょうか。
news.web.nhk/newsweb/na/nd-2…
#nhk_video
2
3
236
🔐 Node.js security updates are now available!
Including the following dependency updates to address public vulnerabilities:
- llhttp on 22.x, 24.x, 26.x
- undici on 24.x, 26.x
- sqlite on 22.x, 24.x
Available for 26.x, 24.x, 22.x Node.js release lines nodejs.org/en/blog/vulnerabi…
M.Sakae retweeted
この界隈をギュらないでください、ガチでお願いします。私の研究を奪わないで。本当に。研究のために何犠牲にしてるのか分からなくなってしまうから。お願いだから。
We sped up materials simulations by 36x with Claude Fable 5.2.
In our early testing, Claude Fable 5.2 sped up DFT and DFPT codepaths in Quantum Espresso by 36x and 9x respectively. It did this while running largely in an autonomous loop, checking its work against our verification sets.
Quantum Espresso is the most widely used open-source codebase for DFT, with roughly 600,000 lines of code written in Fortran.
Thread 🧵
M.Sakae retweeted
NVIDIA GR00T 2 の概要|npaka @npaka123 note.com/npaka/n/n65d91f2bbb…
Fable 5.2 shows substantially improved skills in controlling physical robots, even on tasks it has never explicitly been trained to do.
On RoboDojo, a benchmark with a simulated two-arm robot, it outperformed other models and most AI systems built to control robots.
M.Sakae retweeted
ローカルLLM利用のクリニックを受診してみた PCマニアの院長が自力で構築、その導入効果は?
itmedia.co.jp/aiplus/article…
M.Sakae retweeted
📢 AIRoA has released its robot foundation model, "AIRoA-0" (A-0)! 📢
A-0 learns from 16,151 hours of manipulation data recorded by people using the YUBI handheld gripper, without robots. It can perform bimanual tasks such as towel folding and small-parts sorting in environments different from those where the data was collected.
We have also verified its responses when a person moves sorting bins during a task, as well as its operation on four platforms equipped with YUBI: Franka, TX-G2, OpenArm, and UR.
📦 Open-Source Assets:
・Pre- & post-trained model weights (Hugging Face)
・Data collection, post-training, and example inference code (GitHub)
・A technical report detailing methods and evaluation benchmarks
Post-training on your own task data can also be completed in under four hours.*
(*Using a single node with eight H200 GPUs.)
Check it out and let us know your feedback!
airoa.org/updates/20261106/8…
#AIRoA #RobotLearning #PhysicalAI #Robotics #VLA
M.Sakae retweeted
🚨 Node.js warns of HIGH-severity security vulnerabilities.
Security fixes are coming November 11 for Node.js 26.x, 24.x, and 22.x.
All three release lines are affected. Production users should take note.
Details: nodejs.org/en/blog/vulnerabi…
M.Sakae retweeted
東大で行った医療向けAIの開発の主にデータ分野について発表しました!
よかったら見てみてください!
youtu.be/fKuw7DEJmMg?si=mWEc…


















