CMU Intelligent Autonomous Manipulation Lab

🤖We are introducing Grounded Task Axes (GTA) — a zero-shot skill transfer framework that enables robots to perform multi-step manipulation tasks on novel objects by generalizing modular controllers grounded through visual foundation models. ✅ No training, no demo, no finetuning — just zero-shot 🔧 Modular task-axis controllers 🎯 Grounded via visual foundation models 🧠 Generalizes across tools & objects 🥄🍳☕🔩 Scraping, pouring, screwing, insertion Find more about our paper: 🌐 iamlab-cmu.github.io/GTA/ 📄 arxiv.org/abs/2505.11680

Big congratulations to Mark for getting the RAL paper accepted! Check out the website and video to discover how to autonomously insert sensors into corn stalks and monitor nutrient levels throughout the growing season.

🏆 Congrats to the award winners at the Dexterous Manipulation Workshop at #RSS2024! 📜 Best Paper: Zilin Si, @firephinx, @fzeyneptemel, @Oliver_Kroemer, @iamlab_cmu 🎙️ Best Presentation: @kenny__shaw, @pathak2206 #RSS2024, #DexterousManipulation

We are at @ICRA2024! We will present our paper “Estimating Material Properties of Interacting Objects Using Sum-GP-UCB” at Conference Center Room-313 starting 16:30

I had a wonderful time visiting @GTrobotics to present my work on "Planning What to Learn". Thank you @MatthewGombolay for inviting. @core_robotics is doing fantastic inter-disciplinary work on robot skill learning, multi-agent RL and HRI. Slides @ tinyurl.com/ykphcwt9

Congratulations to DR. Mohit Sharma (@m0hitsharma) on successfully completing your Ph.D. thesis on Representation Reuse for Learning Robust Robot Manipulations! It has been such a pleasure to be your advisor of the last years. @CMU_Robotics

How can we use AI & robotics for sustainable agriculture? 🤖🌽 We present Towards Autonomous Crop Monitoring - inserting sensors in corn stalks to monitor plant nutrient over the growth cycle. See paper, additional documentation at kantor-lab.github.io/cornbot 🧵👇 (1/7)

This is happening now at poster #2!

DELTAHANDS is a synergistic dexterous hand framework that uses soft Delta robots. Delta hands are low-cost, easy to build & control, and highly configurable for dexterous manipulation. Check out our paper (arxiv.org/abs/2310.05266) or site (sites.google.com/view/deltahand…) for more info!

Come visit our poster tomorrow @corl_conf on causal skill learning for manipulation!

Curious how causality can help learn interpretable and robust skills for manipulation? Please visit our work on SCALE tomorrow (Nov 9) at CoRL Poster Session 5 at 2:45pm ET! A thread 🧵 (1/4) #CoRL2023 Website: sites.google.com/view/scale-cau… @corl_conf @CMU_Robotics @iamlab_cmu

If you are at #CoRL2023 @corl_conf, come chat with us at our poster at the LangRob@CoRL2023 Workshop or on Nov 8, 5:00PM ET @Poster 4 Website: mohitsharma0690.github.io/multi-res-real… Paper: openreview.net/forum?id=WuBv9…

How can we leverage large pretrained VLMs for precise and dynamic manipulation tasks in real time? Presenting MResT🤖 framework for real-time control using sensing at various spatial and temporal resolutions Come say hi @corl_conf #CoRL2023 Today @ LangRob and Nov. 8 poster 4

@make_your_maark @CMU_Robotics @AII4RA @farm_ng_inc @SCSatCMU @JmpBell Great work! Looking forward to seeing the video.

Have you wondered how robotics & AI could be used in agriculture 🌽🤖? Last week, we deployed our robot in Iowa to insert micro-sensors into stalks to monitor crop health. Stay-tuned for the highlight video of our robot in action! @CMU_Robotics @AII4RA @farm_ng_inc @SCSatCMU

Super happy to share Code as Policies received the Outstanding Paper Award in Robot Learning #ICRA2023 @ieee_ras_icra !! Big thank yous to my collaborators @wenlong_huang @xf1280 @sippeyxp @hausman_k @brian_ichter @peteflorence @andyzengtweets y'all rock 🎉🎉

Hello all, I'm very excited to present my work on "Linear Delta Arrays for Compliant Dexterous Distributed Manipulation" that I worked on during my Masters in Robotics at @CMU_Robotics tomorrow at @ieee_ras_icra Pod 44 at 9 - 10:40 AM British Summer Time!

🤖🦾 Recent research shows that pretrained vision representations (PVRs) can be useful for robot manipulation tasks. Do these PVRs scale to complex manipulations? In our #ICLR paper we explore this and uncover some fascinating findings @iclr_conf arxiv: arxiv.org/abs/2304.06600

How do you adapt dynamics models to new scenarios? @PeterMitrano, in collab. with @alex_lagrassa and @OliverKroemer, shows that the key is to FOCUS on where source and target dynamics are similar. This does better than learning on all the data! sites.google.com/view/focused-a… #ICRA2023