Yifei Simon Shao

PhD @ UPenn GRASP , human robot collaboration , neurosymbolic planning

[lily zhang @lily_gpupoor]

Introducing 𝐒𝐩𝐞𝐜𝐮𝐥𝐚𝐭𝐢𝐯𝐞 𝐃𝐞𝐜𝐨𝐝𝐢𝐧𝐠: 𝐇𝐨𝐰 𝐈𝐭 𝐄𝐯𝐨𝐥𝐯𝐞𝐝, 𝐖𝐡𝐞𝐧 𝐈𝐭 𝐒𝐭𝐚𝐲𝐬 𝐋𝐨𝐬𝐬𝐥𝐞𝐬𝐬, 𝐚𝐧𝐝 𝐖𝐡𝐚𝐭'𝐬 𝐍𝐞𝐱𝐭. An interactive tutorial @Madisonkanna and I built for the NeurIPS Education Track.

Every LLM you use generates one token at a time. Autoregressive decoding is the bottleneck for inference. Speculative decoding accelerates this, and today it runs under nearly every hosted LLM. It is a cornerstone topic to learn in the LLM stack.

Blog: https://t.co/xE2OxuosgE

[Haoru Xue @HaoruXue]

🚨 Call for live demo / paper at CoRL 2026 𝗔𝗴𝗲𝗻𝘁𝗶𝗰 𝗥𝗼𝗯𝗼𝘁𝗶𝗰𝘀 𝗪𝗼𝗿𝗸𝘀𝗵𝗼𝗽! No paper required for demo track. Due Sep 27.

🔮 The real magical power of agent is live interactive demo with the audience, which is why we provide all-out support:

- Robot, compute, and API are ready for you in the conference room

- Custom robot / sim demo also welcomed

- Just submit a video proof - friendly to industrial participants

https://t.co/F2L9Q05PAK

Paper submission also welcomed!

We also have an amazing speaker lineup from CMU, UC Berkeley, DeepMind, NVIDIA, and Tencent.

[Yifei Simon Shao @YifeiRobotics]

Excited to be part of the team!

[Dyna Robotics @DynaRobotics]

Today we are introducing Dyna-2, a world-action model pre-trained on one million hours of human video. At this scale, for the first time, we discovered several new scaling laws:

• world-action models exhibit scaling law on human data across four orders of magnitude, from 1000 to 1,000,000 hours,
• this human data scaling law implied a scaling law on never seen robot data,
• both data and objective matter; world modeling and scaling on video data are essential for cross-embodiment scaling transfer to emerge

🧵

[Yifei Simon Shao @YifeiRobotics]

@BoWang87 Our frameworks enables these agents to deploy and share skills in between themselves https://t.co/n8AIpv9xRg

[Yifei Simon Shao @YifeiRobotics]

@irvinxyz Super interesting! We did the same thing but allowed robots to share learned skills with each other on our platform! https://t.co/qNZYmIqPYa lets connect

[Yifei Simon Shao @YifeiRobotics]

What do "robot skills" look like when humans and AI build them together? We built 3 essential layers:

🧩 Skills: A community-built library. Browse, wish for a skill, or build it with an agent.
🤖 Agent Server: Ensures safety and hardware access. It manages time, executes safely, and rewinds movements if needed.
⚙️ Services: Hardware drivers and shared models that Service Agents add automatically upon request.

[Yifei Simon Shao @YifeiRobotics]

Call for community effort 🤖! Bring the Lobster 🦞 on robots! Our work Tidybot Universe 🌟, allows AI agents to autonomously iterate on real robotics hardware.

We built the infrastructure for displaying and sharing the learned skills across agents; and for hosting common services to be access across agents, starting with the Tidybot.

Join us today https://t.co/McPAyTLNWw

[Yifei Simon Shao @YifeiRobotics]

@andromeda74356 @chris_j_paxton edu version is not gonna be 30k though!

[lily zhang @lily_gpupoor]

Rising Star Speaker @YifeiRobotics spoke about
𝘎𝘦𝘯𝘦𝘳𝘢𝘭𝘪𝘴𝘵 𝘙𝘰𝘣𝘰𝘵𝘴, 𝘓𝘪𝘬𝘦 𝘓𝘓𝘔𝘴, 𝘕𝘦𝘦𝘥 𝘢 𝘓𝘪𝘵𝘵𝘭𝘦 𝘐𝘯𝘧𝘦𝘳𝘦𝘯𝘤𝘦-𝘛𝘪𝘮𝘦 𝘏𝘶𝘮𝘢𝘯 𝘐𝘯𝘱𝘶𝘵. Having human-in-the-loop is important!

[lily zhang @lily_gpupoor]

Join us on Oct 24th at #IROS2025 RoboGen Workshop! 🤩 We will discuss, Building toward embodied AGI: solving the data bottleneck. We got an exciting line of speakers from academia and industry! Link in thread.
@siyuanhuang95 @ShenlongWang Peter KT Yu
@yuewang314 Ruigang Yang @wu_chenfei @zhijianliu_
Yifei Shao @KungPou40786 @IROS2025

[Yifei Simon Shao @YifeiRobotics]

@xavierpuigf @AIatMeta I am a 4th year PhD work on human-robot task-level collaboration. I wonder what Meta's priorities are this summer!