I am honored and grateful to receive an 𝗡𝗦𝗙 𝗖𝗔𝗥𝗘𝗘𝗥 award “𝘚𝘩𝘢𝘱𝘪𝘯𝘨 𝘌𝘮𝘣𝘰𝘥𝘪𝘦𝘥 𝘐𝘯𝘵𝘦𝘭𝘭𝘪𝘨𝘦𝘯𝘤𝘦 𝘛𝘩𝘳𝘰𝘶𝘨𝘩 𝘓𝘢𝘯𝘨𝘶𝘢𝘨𝘦-𝘎𝘶𝘪𝘥𝘦𝘥 𝘐𝘯𝘵𝘳𝘰𝘴𝘱𝘦𝘤𝘵𝘪𝘰𝘯.” 🎉 This project will support our work in the 𝗣𝗟𝗔𝗡 𝗟𝗮𝗯 toward more reliable, self-improving embodied AI systems. I am deeply thankful to @NSF, my students, whose creativity and hard work inspire me, as well as my collaborators, mentors, and @iSchoolUI colleagues for their support. I am also grateful to my former academic home @VT_CS where this work began to take shape. Excited for the work ahead! Learn more about our lab 👉 plan-lab.github.io #NSFCAREER #NSFfunded #EmbodiedAI #MultimodalAI #ResponsibleAI #iSchoolUI #PLANLab

Two posters from my group and collaborators at #ICML2026 🇰🇷 📍 LaTtE-Flow (~6x faster inference): Tue 7/7, 10:30 AM – 12:15 PM KST, Hall A #2612 📍 UniDFlow (3x fewer params, SOTA on generation, understanding, editing): Wed 7/8, 10:30 AM – 12:15 PM KST, Hall A #3817 Will not attend @icmlconf this year, but @onkar_s_13 and @dhjang10 will be there! Stop by 👋

🤖 🧭 How does a robot know how far along it is in a long task? Most VLA policies cannot explicitly track task progress, which is why they are often brittle in long-horizon, multi-step tasks. PALM (#CVPR2026) makes the policy progress-aware, coupling future affordance prediction with continuous progress estimation in a closed loop. Our code and model checkpoints are released 👇

Small note: the calendar is set up in 30-minute slots. For a longer meeting, please feel free to book two consecutive slots. You are also welcome to schedule follow-up conversations as needed.

It's been an incredible year, with wonderful collaborators and a CAREER award I'm grateful for. I got a lot of feedback on mine, so I want to do the same for others. Opening office hours for faculty writing/thinking about #NSFCAREER proposals. Bring a draft, questions, or just chat. calendly.com/lourent2-illino… Happy to share tips and give honest feedback.

Can I be a Googler for a day? 🌀 Soaking up AI Systems DevLabs 2026 🤩

Eman Abdelrahman, Ph.D. student @SanghaniCtrVT and Spring 2026 Ph.D. grad Makanjuola Ogunleye attended #CVPR2026 w/advisor @Ismini_L, where they teamed up to present their work on “3-VCD: Hallucination Mitigation in 3D-LLM Embodied Agents through Visual Contrastive Decoding.”

Check out our #ECCV2026 work on 🎥 Interactive graph-controllable video generation 🧩 Semantically grounded part-level 3D generation 🚶 Children's gait video understanding Huge congratulations to @VedantRShah, @yu_tianjiao, @SivanShen, Boyi Li, @IrohXu, and all coauthors. Excited to share more soon.

How to fix this? Are per-author submission caps and stricter rejects the solution? 🤔

Hiring and tenure reward count. And now the pressure peaks early, on the student and postdoc job market. Add field growth, conference deadlines that batch everyone into the same few venues, resubmission churn, and now LLMs lowering the barriers of production.

In every cycle and every conference, the submission numbers climb, and I keep asking what we're optimizing for: more papers, or better science? I don't think those are the same thing anymore...

Nothing keeps you humble like a "after considering the rebuttal and other reviews, I maintain my score" right after a good week of #ECCV2026 acceptances 😂 Respect to everyone doing the unpaid reviewing ... and to the rebuttal genre, truly an art form!

This VLA framework is part of our broader work on language-guided introspective embodied agents. 💻 Code: github.com/PLAN-Lab/PALM 📄 Paper: arxiv.org/abs/2601.07060 🌐 Project: plan-lab.github.io/projects/… 🤗 Model: huggingface.co/PLAN-Lab/PALM 🧵4/4

PALM achieves state-of-the-art performance on CALVIN ABC-D and LIBERO-LONG, with robust real-world generalization. 🧵3/4

Long-horizon robot manipulation is challenging. A robot has to chain many steps together (pick this, move there, place it, repeat) and it easily loses track of where it is in the task. PALM tackles this in two ways: 🎯Affordance reasoning: PALM predicts the cues that matter for interaction, such as which objects are relevant, where to make contact, where to place things, and how to move. 📈Progress awareness: PALM estimates how far along it is in each subtask, so the policy knows what it has done and what comes next. 🧵2/4

🤖 Code and model checkpoints are out for PALM, our #CVPR2026 work on long-horizon robot manipulation. Work led by Yuanzhe Liu @ PLAN Lab👏 🧵1/4👇

Went to bed early for the #ECCV2026 AoE release, but ended up wide awake at midnight reading emails ... Worth it though! 3 papers accepted at #ECCV2026 🎉 Congratulations to all collaborators, especially @yu_tianjiao @SivanShen and @VedantRShah for leading these works! 👏 👏🏼👏🏾 More details soon!

Checking the @eccvconf Openreview portal has become my cardio😮‍💨 #ECCV2026

This project is a collaboration between PLAN Lab @ UIUC, Sony Research India, Google, UT Austin, and IIT Roorkee. Huge thanks to the full team and to the TPU Research Cloud program for the resources! PyTorch and JAX versions to be released soon 😊 Project page: plan-lab.github.io/projects/… 🧵6/6

Across benchmarks, UniDFlow shows strong performance on both multimodal understanding and generation, while also generalizing to tasks such as inpainting and multi-subject editing. 🧵5/6