Grounding language in robotic affordances @GoogleResearch
Grounding language in robotic affordances  @GoogleResearch
Uploaded August 2022 | Updated September 2026, 2 weeks ago
Welcome to ResearchBytes - a series that converts research publications into byte-sized content.

In this first episode, we share how Google Research and Everyday Robots worked together to combine some of the latest, cutting-edge research in natural language processing and robotics to help power tomorrow’s helpful, human-centered robots, while simultaneously improving advanced language models like PaLM as they interact with the real world.

Resources:
Read the Palm-SayCan blog post → https://goo.gle/3dGbCaO
Read the paper → https://goo.gle/3P2cbc8
Learn how it works → https://goo.gle/3QppUef

Chapters:

0:00 - Intro
0:37 - What questions will we answer in this video?
0:57- What’s the challenge?
1:57 - Moravec’s paradox
2:29 - What did we do?
3:16 - How did we do it?
4:05 - What did we observe?
5:32 - What’s next and how can you learn more?


Watch more episodes of ResearchBytes → https://goo.gle/ResearchBytes
Subscribe to the Google Research Channel → https://goo.gle/GoogleResearch

#ResearchBytes
Grounding language in robotic affordancesWhy Robots Struggle to Pick Up Small ObjectsHow AI is Improving Global Access to Reliable Flood ForecastsPredicting protein function with ProtNLMFighting extreme heat in urban environmentsResearch@ Bangalore: A conversation with Jeff DeanFrom satellite imagery to queryable insights l Google Earth AIHow we built two new brain maps using Google AIBuilding an ML model to accurately estimate the reflectivity of roofsResearch@ NYC: A Fireside Chat with Corinna Cortes & Katherine ChouAfrican Language ExpansionScaling Language-Image Learning in 100 Languages with PaLI
Google Research |

Grounding language in robotic affordances

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER