💥Introducing FACTR 2, learning external force sensing on commodity robot arms without needing dedicated sensors. We show that learned force signals enable force-feedback teleop on low-cost arms and improve BC policies. FACTR 2 consists of: 1. Neural External Torque (NEXT): learns external forces without needing dedicated force sensors. 2. Force-Informed Re-Sampling Training (FIRST): uses the learned force signal to identify task-critical regions and upsample them during training. w/ @StevenOh_ @_tonytao_ 🧵(1/N)

Two years ago we released VQAScore with a simple idea: ask a VLM "does this image show {prompt}?" and use P(Yes) as the score. It has since become the go-to evaluation metric and reward model for visual generation, replacing CLIPScore across the field. Adopted by Google DeepMind (Imagen 3 & 4), NVIDIA, ByteDance, and other frontier labs. Our open-source model has 2M downloads on HuggingFace. Today: a major upgrade. VQAScore now supports text-to-video evaluation using 20 SOTA VLMs including GPT, Gemini, and Qwen, capturing generation accuracy for prompts like "a shallow depth of field shot rack focusing from a foreground crumpet to a dog entering the background, catching it at the focus shift." As VLMs get stronger, VQAScore gets stronger. For free. 📄 Paper: arxiv.org/abs/2404.01291 💻 Code: github.com/linzhiqiu/t2v_met… Thanks to @chancharikm for driving this upgrade, and @gneubig and @RamananDeva.

Congratulations, Deepak! 👏 Learn more about the award and Deepak's work at CMU and beyond: ri.cmu.edu/pathak-receives-2…

We had amazing Robotics Institute representation at #ICRA2026 in Vienna and #CVPR2026 Denver! 🤖⚙️ Congratulations to all on their hard work! 💪 #TartanProud #CMUrobotics

Apply for the Foundations of Robotic Interactions online graduate certificate program! Next Start Date: August 2026 ⚡ Priority Deadline – June 10, 2026 ✅ Final Deadline – July 29, 2026 Learn more: cmu.edu/online/foundations-r…

From a single demonstration, can a robot learn robust, dynamic bimanual manipulation? We introduce 🧬𝗣𝗚𝗗𝗚, a curation-guided data generation framework that expands one demo into a compact dataset that enables robust BC policy learning. website: cunxid.github.io/PGDG/ 🧵1/6

Excited to share that IA-TIGRIS, a major part of my PhD work, has been published in IEEE Transactions on Robotics! Hundreds of flights and extensive validation toward practical online informative path planning for real robots. ia-tigris.github.io

As the inaugural Bachelor of Science in Robotics graduate, Beverly (Bev) Da Costa not only moved through a brand-new curriculum but also helped define it #TartanProud #CMUgrad 🎓 🎉 Hear from Bev in a Q A where she reflects on her BSR experience: ri.cmu.edu/work-hard-and-dre…

Congratulations, Parv! #CMUgrad #CMUrobotics 2026 grads: tag us in your graduation posts! We want to celebrate with you! #TartanProud

Congrats to @LeCARLab's 2026 #CMUGrad!

Excited Carnegie Mellon students had the opportunity to meet with Jensen Huang, @nvidia Founder and CEO. Jensen engaged firsthand with the cutting-edge research being conducted by CMU students. There was a snake robot, a robotic arm — even a robobuggy.

Congratulations to our Founder and CEO Jensen Huang on receiving an honorary Doctor of Science and Technology degree from @CarnegieMellon, as well as the CMU Class of 2026 graduates. 🎓

This morning, Jensen Huang, Founder and CEO of @NVIDIA, met with Carnegie Mellon students ahead of Commencement. The visit created an opportunity for students to connect and share their work. Watch his 2026 remarks: cmu.edu/commencement/

Time to celebrate the Class of 2026! 🎓❤️ Grads, families and friends: Share your photos from Carnegie Mellon commencement with the #CMUGrad hashtag!

World2Rules learns patterns of risky aircraft behavior from real airport operations & incident data, then explains warnings in terms humans can easily understand ✈️ 🧑‍✈️ It's designed to work alongside existing prediction systems to enhance safety! ri.cmu.edu/?p=151221&preview…

Leslie Kaelbling, Panasonic Professor of Computer Science & Engineering @MIT, gave the last seminar of the semester! You can find her talk, "The Role of Rationality in Modern Robotics", on our YT channel along with a playlist of all other seminars ⬇️ bit.ly/4wbW9TU

May the fourth be with you. #StarWarsDay

Register for the next CMU Vision-Language-Navigation challenge, which will conclude with a workshop at #IROS2026! Teams will develop systems that allow robots to interpret natural-language instructions and navigate unfamiliar environments. Learn more ⬇️ ri.cmu.edu/next-phase-of-vis…

Carnegie Mellon robots debuted their skills on the #NFLDraft stage! 🤖 'Simon' from CMU's Robomechanics Lab and 'Maggie' from CMU spinout HEBI Robotics were drafted to showcase Pittsburgh's innovation in #robotics.

Before AI can generate professional videos, it needs to see like a professional. We spent a year with 100 content creators teaching AI to describe video like a filmmaker would. Introducing CHAI: Critique-based Human-AI Oversight for Building a Precise Video Language [CVPR'26 Highlight, Top 3%]. Try prompting a video generator for a dolly zoom, dutch angle, point of view, or camera roll. Most fall back to the same bland defaults: a push-in, a level shot, a third-person view. Why? These techniques require a language of cinema that current models rarely speak. We built that language: 1️⃣ Precise specification: 5-aspect structured captions co-designed with professional cinematographers covering subject, scene, motion, spatial, and camera dynamics 2️⃣ Scalable oversight: LLMs draft captions, humans critique what's wrong and how to fix it 3️⃣ Post-training recipes: Qwen3-VL-8B surpasses Gemini-3.1 and GPT-5 4️⃣ Video generation: fine-tuned Wan follows 400-word cinematic prompts with precise control Here's how each works 🧵 Work led by CMU and Harvard with @chancharikm, @du_yilun, and @RamananDeva. 📄 Paper: huggingface.co/papers/2604.2… 🌐 Site: linzhiqiu.github.io/papers/c…