Gemini 4 Argon sank seven cornhole shots on Apollo
Google says it showed its model a video of a human playing cornhole, and the Apptronik Apollo humanoid then hit seven straight. Takahiro Miki, who ran the demo, posted the blow by blow, including a throw that fell 85 cm short because the grip slipped.
Shown this week
Google posts the cornhole run, but the full thread shows way more than seven throws
Google says Gemini 4 Argon watched a video of a human playing cornhole, built its own simulation code to practice, and then its Apptronik Apollo humanoid sank seven straight shots.
Large models really are changing a lot of the assumptions about how intelligence for robotics is going to work
AlphaNIST solves all 20 NIST assembly tasks at 98.2% in sim
Octi Zhang released AlphaNIST, a single policy trained from scratch on a binary success or failure reward that he says solves all 20 NIST assembly tasks at a 98.2% success rate in simulation. He says it can also finish 10 different tasks back to back while getting the reward only once, at the end.
First promising result for sim2real in manipulation
Out in the world
Physical Intelligence opens its Pi API, from strawberries to servers
Physical Intelligence put the Pi API into beta, saying it has already partnered with companies to adapt its models to their robots, from harvesting strawberries to assembling servers. Chelsea Finn says many people have been fine-tuning the pi models on their own robots already and the team wants to expand testing a lot.