🛫 I’m heading to Pittsburgh for IROS 2026 🤖 9/27–10/2. Introducing
🥂
@saturdayrobotic × #
IROS2026#: Saturday Robotics Research Night
📍 Pittsburgh | Sept 28, 5:30–9:30 PM
🎟️ RSVP:
Lightning talks from researchers at
@Caltech,
@RhodaAI ,
@XSquareRobot,
@CAS__Science,
@UCBerkeley, and
@HarryXu12 (
@PokeBot_Global), covering:
• humanoid safety + RL
• in-context learning for robot data
• scalable sim evaluation
• 3D VLA + memory + inference-time learning
• hindsight reward editing
• GPT-6 Astra & the future of robotics
The interesting part isn’t just the demos.
It’s figuring out what actually scales—and what doesn’t.
A few questions to discuss at Saturday Robotics IROS research night:
→ Is robot scaling really about more data—or better experience per sample?
→ Can in-context learning turn robot data from static demonstrations into reusable experience?
→ Will world models become the data engine for robotics?
→ What should robots remember over long horizons?
→ And what does scaling not fix—representation, architecture, reward, evaluation?
If you’re working on robot foundation models, world models, in-context learning, robot data, sim2real / Real2Sim2Real—or just trying to answer “how do we actually scale robot learning?”—let’s grab coffee ☕️.
See you in Pittsburgh. 👋
@aurorafeng_01 @michellelsun