Geek+

Headlines · 3

1. 5d
   Geek+ introduces Gravity Control for robot runtime error correction

   Geek+ recently introduced Gravity Control under its Gravity embodied-intelligence framework. The report says the method uses causal interventions on internal model features so natural-language prompts such as “grab the bread deeper” can correct local action deviations without updating the original policy weights or adding demonstration data. In reported physical-robot tests on out-of-distribution manipulation tasks, overall success rose from 17.5% to 87.5%, while the sideways-bread task improved from 5% to 95%. The method is intended to correct execution deviations when the robot already possesses the relevant skill, rather than teach it a new task.

   新智元 · 🔥 0
2. 5d
   China's Embodied AI Investment Cools as IPO Prospects Weaken

   36Kr reports that Chinese embodied AI financing cooled sharply, citing a 40% month-over-month decline in September, fewer projects reaching investment committees and tighter funding criteria. The report attributes the shift to weaker public-market returns, stalled IPO expectations, concerns that general-purpose models could reduce the value of specialized robot models and signs of speculative valuation practices. Some startups are restructuring or seeking smaller rounds, while investors are placing greater emphasis on proven demand, commercial deployment and industrial resources.

   36氪 人工智能 · 🔥 0
3. Oct 21
   Vertex Ventures' Xia Zhijin: embodied AI won't see a winner-take-all endgame

   In an interview with Cyzone during the 2026 DEMO CHINA event, Vertex Ventures China managing partner Xia Zhijin argued that embodied AI will not end in a winner-take-all market and may end up less concentrated than the auto industry, with many scenario-specific leaders rather than one general-purpose company covering every use case. He said deployment will be gradual, with robots of different forms targeting different domains over the next few years before capabilities become more general beyond five years. Xia identified data as the biggest bottleneck for embodied foundation models, citing high cost and low information density compared with internet text, and said data companies should layer cleaning, labeling and downstream use on top of collection to avoid price wars and homogenization; he added that simulated data can help generalization and long-tail corner cases across pretraining and later training stages.

   创业邦 科技 · 🔥 0

Experience and discussion from the community

Share my Geek+ experienceAsk about Geek+