Assistant Prof. @GMUCompSci. Research at the intersection of Robotics, Planning, and Machine Learning to make robot planning more comfortable with uncertainty.
Our new paper develops robots that donโt just complete tasks: they anticipate how their actions impact what comes next.
Example: when putting objects away, organizing them neatly isnโt just aestheticโit makes future retrieval faster and easier. ๐๐งต๐
We demonstrate anticipatory TAMP on a real Fetch mobile manipulator.
Compared to myopic planning, anticipation reduces the number of actions and total execution time over task sequences.
We're excited to announce the third workshop on LEAP: Learning Effective Abstractions for Planning, to be held at #CoRL2025@corl_conf!
Early submission deadline: Aug 12
Late submission deadline: Sep 5
Website link below ๐
Most student authors (or other first time BibTeX users) don't realize that extra curly brackets around an {ACRONYM} are needed to preserve capitalization. It's *such* a small thing, but it makes clear the bibliography hasn't been proofread (and doesn't look good in a thesis ๐จโ๐)
On my way to Baltimore to attend the NSF FRR/NRI PI meeting and excited to share some recent work and talk robotics for the next couple days ๐ค. Be sure to say hello if you see me!
Amazing how quickly a huge project can go from high-effort-not-nearly-done to *done*. Been working on a grant for weeks and putting time towards it nearly every day and then suddenly ๐ every section is drafted, every figure made, no more todo items... ๐ซ
With all the acceptance news RE @ieee_ras_icra, my lab got... 0 papers accepted!๐ A reminder that not all good papers get accepted & rejection can be an opportunity for further improvement. Still proud of my lab's work and excited to share (just a bit later than expected)!
Just got back from a great visit to @ASU (thanks to @nakulgopalan for hosting me!) where I talked about some of my lab's ongoing work in making long-horizon planning in partially revealed environments more performant and reliable.
Credit goes to my student and first author @abpaudel. Stop by his talk or poster at #IROS2023 [WeBT17] or, if you're not in Detroit, read our blog post: https://t.co/8p615IpO2b See our paper ( https://t.co/Qqfuce0Qtu ) for more experiments and details.
Very excited to present our work: Data-Efficient Policy Selection for Navigation in Partial Maps via Subgoal-Based Abstraction! We enable a learning-guided robot deployed in partially-mapped environments to quickly/reliably select the best of a set of nav policies #IROS2023 ๐ค๐งต
Our approach converges *much* more quickly resulting in lower average navigation cost and cumulative regret in simulated maze and office-like environments.