yesterday i advised self-censorship to a undergrad student im helping apply for grfp re: their research & personal motivations. i think they've got as good a shot as any other. i think im making the right call, but it feels deeply wrong and capitulatory. any thoughts (esp from reviewers)?

Last day at the GISS building (Wed. May 28th). 🥲

NSF GRFPs terminated for those in graduate school and already attending Harvard, too.

Me and @wxie.bsky.social

for my area exam last month i reviewed 25 touch-based IL works from 2024/5, taxonomized across tasks, sensor type, model architecture, and more (up now at arxiv.org/abs/2504.11827). here's a perhaps intuitive plot of policy action space--the in-figure citations are clickable in the pdf/html!

I'm giving a lecture on language model debiasing to my undergrad NLP course on Friday but I'm not super up to date on the research. Does anyone have any suggestions for papers/topics to cover?

taking name suggestions…

how i'm feeling now

pleasantly surprised at how easy it was to setup rerun data viz for both my RLDS dataset and its raw data

reading a pratchett speech from 1985 on the wizard/witch dichotomy in fantasy: ansible.uk/misc/tpspeec... "The classical witch, on the other hand, with her often malevolent interest in the small beer of human affairs, is everything we fear only too well that we would in fact become."

i read five papers on tactile robot policies and wrote about them: medium.com/correll-lab/... the first step of me writing my prelim and i hope to have some more interesting synthesis soon. this is in no way a comprehensive review and i welcome recommendations of papers to read!

hyperlocal politics warning/enormously hot take—the chinatown superjail is not tenable, for so many reasons. i would rather the city continue failing to run rikers than nuke the neighborhood. this will be a bigger deal than the proposed shsat reform, because it is actually happening

is anyone exploring (or know of work on) motion event typology for language robot models? seems like this is the kind of underlying structure we can/should leverage across embodiments/contexts. stuck on this passing thought and don't know the right search terms, so sorry if this is uninformed

Hi, I'm a Professor of robotics at CU, writing about robotics on my Medium blog medium.com/@nikolaus.co... and our lab's work on medium.com/correll-lab, and looking forward to amplify robotic news on BlueSky.

my corl postmortem for the void, very loosely 1/? quoting Jitendra Malik, it took us ~500 million years to develop human-level sensorimotor capabilities and ~100,000 years to develop language. to riff off that, is the robot data scaling problem "simply" 5000x the semantics problem?