Twitter Beğeni Hilesi

The first #PaperILike is "Pengi: an implementation of a theory of activity" (Agre & Chapman, 1987). This paper points to a kind of improvisational decision-making that I still don't think we've figured out. Also has a great critique of planning. PDF: cdn.aaai.org/AAAI/1987/AAAI…

I'll be at #RSS2024 next week to present our "planning to learn" work (ees.csail.mit.edu). Prospective PhD students: feel free to email me if you want to chat about robotics at Princeton.

Excited to be at #RSS2024 next week! Stop by our talk + poster during the Planning session on Tuesday for fun videos of robots learning in the wild and handling adversarial interventions from unfriendly gremlins 👿 (we promise no gremlins will be at the presentation!)

This week's #PaperILike is "Finding Options That Minimize Planning Time" (Jinnai et al., ICML 2019). This paper has a unique & compelling answer to the question: how should we discover high-level action abstractions (and why)? PDF: proceedings.mlr.press/v97/jinnai19a/…

Exciting to see TAMP playing such a prominent role in Dieter Fox's plenary talk at #RSS2024! A bonus #PaperILike for this week is one he highlighted: "Human-in-the-Loop Task and Motion Planning for Imitation Learning" (Mandlekar et al. 2023) PDF: arxiv.org/pdf/2310.16014

This week's #PaperILike is "Learning Symbolic Models of Stochastic Domains" (Pasula, Luke Zettlemoyer, Kaelbling, JAIR 2005). This was hugely influential in my PhD. The scope is incredible (you think they're done; then they add concept learning!) PDF: people.csail.mit.edu/lpk/papers/zpk…

I'm a big fan of Gregory Stein 's "anticipatory planning" ideas and proud to be a small part of this latest work: arxiv.org/abs/2407.13694

I am excited to share that I will join the ECE department at Stony Brook University SBU Dept. of Electrical and Computer Engineering Stony Brook University as an Assistant Professor starting this fall. I will be recruiting PhD students during this application cycle to work on robot learning!

This week's #PaperILike is "ORGANA: A Robotic Assistant for Automated Chemistry Experimentation and Characterization" (Darvish et al., 2024). TAMP, LLMs, and automated chemistry? I'm in! PDF: arxiv.org/pdf/2401.06949

Six years ago I wrote a post on "Lessons from My First Two Years of AI Research." I got a lot of nice feedback and messages from people over the years saying that it helped. Now that I've finished another career chapter, I've added a few more lessons: tomsilver.github.io/blog/2024/less…

This week's #PaperILike is "Discovery of Complex Behaviors through Contact-Invariant Optimization" (Igor Mordatch et al., 2012). The behaviors synthesized are still among the coolest I've ever seen. An elegant idea, too. Video: youtube.com/watch?v=mhr_jt… PDF: homes.cs.washington.edu/~todorov/paper…

How do robots understand natural language? #IJCAI2024 survey paper on robotic language grounding We situated papers into a spectrum w/ two poles, grounding language to symbols and high-dimensional embeddings. We discussed tradeoffs, open problems & exciting future directions!

Nice article from MIT News on our recent work! news.mit.edu/2024/helping-r…

🚀 Excited to announce the launch of the Simulation Phase (Phase 1) of our challenge! Register by Sep 8 & submit by Oct 31. 🗓️ Participate in one or both of the tracks: Track 1: Fixed-base Manipulation for Robot-assisted Dressing Track 2: Mobile Manipulation for

This week's #PaperILike is "Learning by Analogy: Formulating and Generalizing Plans from Past Experience" (Carbonell, 1982). Ambitious and ahead of its time. Analogical reasoning is under-utilized in modern AI, imo. PDF: apps.dtic.mil/sti/tr/pdf/ADA…

This week's #PaperILike is "Inductive Logic Programming At 30: A New Introduction" (Andrew Cropper & Sebastijan Dumancic, JAIR 2022). When I first started reading about ILP, I really wanted an explanation in modern ML terms. This does that & a lot more! PDF: arxiv.org/pdf/2008.07912

This week's #PaperILike is "CORN: Contact-based Object Representation for Nonprehensile Manipulation of General Unseen Objects" (Cho et al., ICLR 2024). The real nonprehensile manipulation videos are pretty remarkable. Video: youtube.com/watch?v=TQE-Wk… PDF: arxiv.org/pdf/2403.10760

This week's #PaperILike is "Learning Tasks From A Single Demonstration" (Atkeson & Schaal, 1997). A title ahead of its time! My favorite part, though, is the thorough & thoughtful discussion section. PDF: cs.cmu.edu/~cga/papers/cg…

We are excited to announce the 2nd Workshop on Learning Effective Abstractions for Planning (#LEAP2024) at Conference on Robot Learning! 🐸 Join us as we explore abstraction learning for robotics with a stellar line-up of speakers. #CoRL2024 Website: leap-workshop.github.io Deadline: Sep 22, 2024

The first LEAP was a ton of fun. I'm sure the second one will be a blast too. Submit and attend!