新浪科技讯 3月10日上午消息,由原华为天才少年稚晖君(彭志辉)创立的智元机器人今日发布,首个通用具身基座大模型GO-1。
据悉,该模型开创性地提出了Vision-Language-Latent-Action (ViLLA) 架构,该架构由VLM(多模态大模型) + MoE(混合专家)组成,其中VLM借助海量互联网图文数据获得通用场景感知和语言理解能力,MoE中的Latent Planner(隐式规划器)借助大量跨本体和人类操作视频数据获得通用的动作理解能力,MoE中的Action Expert(动作专家)借助百万真机数据获得精细的动作执行能力,三者环环相扣,实现了可以利用人类视频学习,完成小样本快速泛化,降低了具身智能门槛,并成功部署到智元多款机器人本体。(文猛)
据悉,相比已有的最优模型,GO-1平均成功率提高了32%(46%->78%)。其中,在执行“Pour Water”(倒水)、“Table Bussing”(清理桌面) 和 “Restock Beverage”(补充饮料) 任务表现尤为突出。
责任编辑:尉旖涵
Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.