The LLM code review bots are fantastic. Sure, the make mistakes that are easily identified but they also catch a lot of sneaky bugs. Really do wish I could use them in academic settings but the cost…

A thought-provoking set of thoughts related to AI-generated research.

I'm curious to see how (or if?) we choose to moderate this new form of research. I'm not sure how I feel about letting the community that generates this research policing themselves 😅, but I think it's a great way to start!

Brom Version 0.4.3-1 is out!

We've formalized a feature that we've been using to develop tests in Drake: the Puppetmaker.

This feature allows you to freely move an object around the scene (like a Puppet). (1/2)

So interesting (but exciting) to see that HJ Terry Suh has started a robotics company with advising from his former PhD Advisor, Russ Tedrake.

www.carbon6robotics.com/leadership

Excited to see what they will build!

#robotics #automation

Glad to see these protected bike lanes going up around Summit & Pavonia in Jersey City. The city becomes easier and easier to ride my bike in every day. 😊 #ISeeYouJC

Brom Version 0.4.1 is out!

This time, we added features for helping with the "objects slipping out of hands" problem.

If you're familiar with how robotics simulators work, this is a problem of how most simulators handle collision geometries (i.e., they use convex hulls). (1/2)

Official start of L4DC’25

My friend recently made this word game! If you want to test your vocabulary (or if you want to expand it, like I do), then check it out:

word-boundary-challenge-edensjasmine.replit.app

I've been making a series of videos called "What the FUNDING" that highlight all the incredible science that federal funding supports. If DOGE just fcked up your NSF grant and you wanna collaborate on a video, lmk? I'm @science_irl on insta & TT

#scicomm 🧪 #nsf

As a previous NSF GRFP Fellow, I've seen its alumni (my peers) go on to:
(i) develop novel tech for US businesses,
(ii) become professors at our nation's top universities, and
(iii) lead programs at various national labs + similar research orgs.

I'm sad that we won't get more of that [1/2]

A little bit of interesting history: in late 60s, early 70s, two PhD programs were established at the University of Michigan: computing and communication sciences (CCS) in LSA and computer, information, and control engineering (CICE) in CoE.

They had classes in automata theory, AI, info theory, algorithms, optimization, control theory, and even social decision making. It had a perfect Cyber-Physical Systems curriculum before CPS was a thing (well cybernetics was a thing): drive.google.com/file/d/12SEj...

The 2024 Turing Prize winners announced this morning: Barto & Sutton for "developing the conceptual and algorithmic foundations of reinforcement learning". Well deserved. Built on 1959 Donald Michie's idea; matchboxes and colored beads learning tic-tac-toe. awards.acm.org/about/2024-t...

Early this morning Firefly's Blue Ghost vehicle soft landed on the Moon. 10 science payloads onboard. www.space.com/the-universe...

A great workshop for senior PhD
students and postdocs interested in pursuing academic positions: nextprof.engin.umich.edu/nextprof-nex...

1/10 We study LLMs empirically, as we do for many natural systems, as distinct from traditional mathematics and formal computer science. That's OK, but don't be confused about how well we understand them. It is more like how we try to understand economics than, say, electrical engineering.

Willems’ behavioral system theory is an amazing way to think about systems in an abstract and unified manner. I have found myself thinking about it in very different contexts, from model reduction, to learning, to abstractions.

brom_drake updates/recap (since I haven't posted about it on this platform yet):

Brom 0.3.0 now includes:
- the DiagramWatcher object,
- the drakeify_my_urdf function, and
- brom_drake Production objects.

Which we think will combine to make using Drake simpler and more fun!