Unverified50% confidenceEventExact time
The DeepMind team tried letting the Gemini 4 model directly control a humanoid robot to complete the bag-tossing game (Cornhole), and it hit the target precisely 7 times in a row.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/16/2026
First Seen
Valid until: 10/30/2026
Sources
Related Entities
Related Claims
UnverifiedOn the real humanoid robot, Gemini 4 adopted a progressive adjustment strategy, observing the landing point of each beanbag and adjusting the release point by a few centimeters until seven consecutive throws went into the hole.78% similarUnverifiedGoogle DeepMind showed that Gemini 4 Argon, after watching videos of humans playing hacky sack, built a simulation environment in MU9 and computed physics and whole-body control policies.75% similarUnverifiedThe Gemini 4 Argon team, using a single human demonstration video and the Mojoku simulator, enabled a humanoid robot to learn the cornhole throwing task without preset programming.72% similarUnverified谷歌联合创始人谢尔盖·布林近几个月敦促核心 AI 员工全力投入 Gemini 研发，今年 4 月 Anthropic 短暂领先时布林在全员大会上要求 DeepMind 加快步伐68% similarUnverifiedGemini Plays Pokemon项目后期去掉了所有人工编写的子智能体，只给模型提供原始工具，让模型自己在游戏过程中构建子智能体和可复用脚本68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1041642API
curl https://kongchang.com/api/v1/knowledge/claims/1041642MCP
get_claim(id=1041642)