@3Dconnexion please make your space mice devices report their serial numbers via HID! This is critical if you have two connected to the same machine and you need to uniquely identify them, for example for robot teleportation. Thanks!
English
Peter Mitrano
1.3K posts
Peter Mitrano
@PeterMitrano
PhD @UMRobotics in robot learning & deformable object manipulation
Cambridge, MA Katılım Nisan 2015
112 Takip Edilen405 Takipçiler
I have several colleagues now who are too young to have ever used ROS 1
😭💀⚰️
English
I suppose this move! If you don't want to support DEI initiatives that's your right, but I'm glad PSF is sticking to their values.
Python Software Foundation@ThePSF
TLDR; The PSF has made the decision to put our community and our shared diversity, equity, and inclusion values ahead of seeking $1.5M in new revenue. Please read and share. pyfound.blogspot.com/2025/10/NSF-fu… 🧵 pyfound.blogspot.com/2025/10/NSF-fu…
English
Peter Mitrano retweetledi
TLDR; The PSF has made the decision to put our community and our shared diversity, equity, and inclusion values ahead of seeking $1.5M in new revenue. Please read and share. pyfound.blogspot.com/2025/10/NSF-fu…
🧵
pyfound.blogspot.com/2025/10/NSF-fu…
English
@pablovelagomez1 @rerundotio Is RRD an open format? I wouldn't trust putting my datasets into a closed source file format. Obviously rerun is currently open source but I don't know if the file type itself is well documented
English
Working on adding a new dataset to the lineup. Ported ego-dex over to @rerundotio
With rerun now stabilizing RRD format between versions (0.23 -> 0.24), this is the perfect time to start encoding all of the datasets I've been using to RRD
1. I'm starting with ego-dex and then adding others, such as HOCAP/Assembly 101
2. Looking to see if it also makes sense to port to webdatasets <-> RRD
3. I've started including visualizing confidence — green (high), yellow (medium), red (low).
More info on Friday
Pablo Vela@pablovelagomez1
Streaming iPhone data in real-time directly to @rerundotio 🚀 The collection process is one of the most frustrating parts of building imitation-learning datasets. I’ve got a little army of sensors—📱 iPhone, iPad, Quest 3—but getting them temporally aligned, spatially aligned, AND seeing real-time feedback while recording is tough. I stumbled on a great library from @wpicakelab called ARFlow. It’s a thin client built on Unity’s ARFoundation that connects over gRPC to a server running Rerun for live data logging. I forked it to: - Log the SLAM translation poses, and - Upgrade rerun to v0.23 for my use case. So far, it works well, but there are still a few hitches: 1. Right now, it’s solid on iPhone and iPad; my Quest 3 client is still slow and not super reliable. 2. I’m using an older ARFlow branch focused on real-time streaming only—no spatial or temporal sync yet. Unity builds for iOS keep failing. 🛠️ 3. Nothing is saved locally to the client, so packet loss is a risk on shaky networks. There’s huge potential in tapping the ubiquitous sensors we carry around every day, and ARFlow is a big step toward making that easy
English
Please share your thoughts in replies or find me and tell me what you think!
English
On the other hand, suction or parallel jaw are simpler, easier to simulate or model. And the hardware is better so we can gather real world data more cheaply and with less time spent fixing/replacing things.
English
Thought provoking question for everyone at ICRA:
Do we need to achieve dexterity anthropomorphic hands before we can achieve dexterity with other grippers?
By dexterity I mean ~adult human level. Any by other grippers I mean parallel jaw, suction, etc. more in thread!
English
I will be at #ICRA2025 with @BerkshireGrey
Excited to see everyone and talk about robots!
English
I'm looking for recommendations for an optical flow model that's up to date and easy to use (unlike mmflow), and fast (<40ms, unlike flowformer)!
English
@AmtrakNECAlerts if you're going to cancel some services you need to make it easier to switch to other trains. They're not full, but your phone and in-person agents say they can't switch us to the next available trains, and conductors are turning people away.
English
@kscottz Maybe a sign to make/improve an official Chinese translation of the docs lol
English
Can someone who practices the black art of SEO tell me why the Chinese fork of the ROS 2 docs has higher search results than our official docs pages?
English
When you hear "the sky is the limit", maybe it's actually "we don't know when it will stop working"
English
@pthangeda_ Wow I didn't think that would work but it seems most papers fit and the results are somewhat correct! I've added some trimming but that part is jank because PDFs are totally inconsistent 💀
English
@PeterMitrano Yup, none of the RAG or embedding search, just making the most out of Gemini's extremely long context window hahaha
English
I thought it would be a fun easy task to automatically go through all the CoRL papers and find all the papers that propose a method using behavior cloning.... I'm finding it surprisingly hard to find a good method?!? Any suggestions?
English
@pthangeda_ I guess the method you're using here isn't using "embedding search" with vector stores, but just grabbing all the pdf text and including it in the prompt? I could see if that's cheaper I suppose... 🤔
English
@PeterMitrano I feel like you are overestimating the cost of LLM APIs. They are crazy cheap! Gemini Flash-8B costs only like $0.03 per million tokens (an 8 page paper is < 50,000 tokens). You can process all ~300 CoRL papers in less than 50 cents lol (and it works great! - I ran a test)
English
@pthangeda_ Huh very interesting.... I am using gpt assistants and quickly racked up $30 in usage costs (my employer pays for it).
English