Yanjiang Guo

Excited to share VLAW: Iterative Co-Improvement of Vision-Language-Action Policy and World Model We explore improving VLA inside a learned world model, and find that the key is to jointly improve VLA and WM! Website: sites.google.com/view/vlaw-arxiv

I am surprised that so many pre-trained knowledge can be preserved with no additional data if you finetune VLA in a proper way! Check this solid work from Suning!

Scientists often make breakthroughs by synthesizing ideas across papers. In our new paper, we ask whether a language model can anticipate this process: given two parent papers, can it generate the core insight of a future paper built on them? 🧵⬇️

LLM post-training used to mean fine-tuning to a downstream task Robotics has been stuck in this setting, needing task-specific fine-tuning for best performance π07 changes this: It works out of the box & outperforms fine-tuned specialists Details: pi.website/pi07

In a true research, you start with some assumption and maybe end up with something else! A lot to learn in this thread :)

i'm so excited to finally share what we've been working on @rhodaai! taking lessons from success in LLMs, we identify autoregressive video generation as a scalable objective for training robot policies. 1/n

For a long time, I was skeptical about action-conditioned video prediction for robotics. Many models look impressive, but once you ask them to handle long-horizon manipulation with real physical interaction, things quickly fall apart (e.g., Genie is amazing but mostly focused on navigation). This project changed my mind. I'm beyond excited to share Interactive World Simulator, a project we have been working on for the past ~1.5 years 🤖 One of the first world models that produces convincing results for long-horizon robotic manipulation involving complex physical interactions, across a diverse range of objects (rigid objects, deformables, ropes, object piles). It directly unlocks scalable data generation for robotic policy training and policy evaluation. Try it yourself (no installation needed): yixuanwang.me/interactive_wo… Play directly with the simulator in your browser. Key Takeaways: 1️⃣ 15 Hz long-horizon action-conditioned video prediction for 10+ minutes on a single RTX 4090 GPU 2️⃣ Visual and dynamic fidelity: people often ask how much sim data equals one real data point. In our experiments, it turns out to be close to one-to-one using the Interactive World Simulator 3️⃣ Stress testing matters: we emphasize interactive stress testing to understand robustness and stability and to build trust in the simulator 4️⃣ The model is trained with only ~6 hours of real-world random interaction data on a single GPU. Imagine what happens if we scale this 1000× or even 1M× Huge credit to @YXWangBot, who led this effort with countless hours of work on data collection, training recipes, and system design. I'm incredibly proud of the work he did here! Enjoy the demos and videos. We also fully open-sourced the codebase for anyone interested in applying this to their own tasks. #Robotics #RobotLearning #WorldModels #EmbodiedAI

@RoboPapers Thank you so much for having us!💕

Pretraining is essential for good performance on a wide variety of robotics tasks, and so most vision-language-action models build off of a vision language model (VLM) trained on a wide variety of image-language data. But how does the choice of VLM translate to downstream robotics performance? Jianke Zhang and @GYanjiang join us to talk about this key part of the robot policy, looking at a wide variety of different VLMs and how they perform. Interestingly, they see that performance on auxiliary tasks like quesiton answering did not lead to downstream improvements in control. To learn more, watch episode 65 of RoboPapers now, with @chris_j_paxton and @DJiafei!

We equipped PI policies with memory! And taught our robots to do long-horizon real world tasks such as preparing the items for a recipe, cooking a grilled cheese and cleaning the kitchen!

We added short-term visual memory + long-term text memory to pi models. 🤖 Enables robots to: - complete tasks up to 15 min long - cook grilled cheese while keeping track of time - adapt in-context Paper & videos: pi.website/memory

Today's state-of-the-art VLAs struggle to generalize zero-shot to new robot embodiments, despite training on extensive multi-embodiment data. We introduce Language-Action Pre-training (LAP) and LAP-3B — the first VLA to achieve substantial zero-shot transfer to unseen real-world robot embodiments, through simply aligning action representation with language. Everything is open-sourced! Try it out on your own robot: 🌐 lap-vla.github.io

Long-tail scenarios remain a major challenge for autonomous driving. Unusual events—like accidents or construction zones—are underrepresented in driving data, yet require semantic and commonsense reasoning grounded in control. We propose SteerVLA, a framework that uses VLM reasoning to steer a driving policy via grounded, fine-grained language instructions. Paper: arxiv.org/abs/2602.08440 Website: steervla.github.io

Huge thanks to our project co-leads @tonyh_lee and @lucy_x_shi, and our advisors @chelseabfinn, @percyliang, and @JianyuChen_THU!

It's also excited to see many recent papers who also practice VLAs inside world models: World-Gymnast: arxiv.org/abs/2602.02454 World-VLA-Loop: arxiv.org/abs/2602.02454 WMPO: arxiv.org/abs/2511.09515 We tried a policy-agnostic style improvement with expressive π0.5 policy!

Excited to share VLAW: Iterative Co-Improvement of Vision-Language-Action Policy and World Model We explore improving VLA inside a learned world model, and find that the key is to jointly improve VLA and WM! Website: sites.google.com/view/vlaw-arxiv

🤖 Can we demonstrate humanoid complex whole-body manipulation skills without a physical robot present? Introducing HuMI: A portable, robot-free interface for learning diverse humanoid manipulation tasks. 📄 arxiv.org/abs/2602.06643 🌐 …noid-manipulation-interface.github.io