R to @GoogleDeepMind: To be genuinely useful in our homes and workplaces, robots need finesse…

TL;DR - Google DeepMind announced Gemini Robotics 2, a robotics model that brings high-dexterity manipulation to multiple hardware platforms and adds multi-robot collaboration. It matters because cross-embodiment control plus robot-to-robot coordination targets the practical dexterity gap blocking robots in homes and workplaces.

• Cross-embodiment: one model controls both five-fingered hands and parallel grippers, rather than being tied to a single robot form factor.
• Claimed fine-manipulation abilities include tying a knot and screwing in a lightbulb, plus complex packing tasks with grippers.
• New multi-robot collaboration lets heterogeneous robots communicate and jointly solve tasks beyond a single robot's reach.
• Announcement-level content only — no benchmarks, success rates, or architecture details are provided in the post.