英国利兹大学和伦敦大学学院研究团队携手开发了一款人工智能（AI）系统，成功让四足机器人获得 类似猫狗等动物对环境的适应能力，并使其能在陌生地形中自主调整步态，实现智能导航。这项研究发 表于新一期《自然·机器智能》期刊。 新AI系统就像给机器人注入了生物本能，其不仅能模仿动物的小跑、奔跑、跳跃等动作，更学会了根 据地形实时决策，就像真正的生命体那样应对未知环境。实验显示，"克拉伦斯"在不平坦的木材上、松 散的木屑上以及杂草丛生的植被间等复杂地形中都能灵活调整步态。 新AI系统实现了三大创新：首次实现了动物运动所有三个关键组成部分——步态过渡策略、步态程序 记忆和自实时调整的集成，实现了从虚拟训练到现实应用的直接转化，以及赋予机器人自主决策能力， 而无需人工干预。 研究团队希望，该技术将在灾害救援、太空探索等领域发挥重要作用。比如在核泄漏等危险环境中，这 类机器人可替代人类执行任务。未来，他们还将开发跳跃、攀爬等更复杂的动态技能，最终目标是打造 像生物一样灵活的人形机器人。 （文章来源：科技日报） 研究团队从自然界获得灵感，将猫、狗、马等四足动物根据需要调整步态以节省能量、保持平衡、应对 威胁的生存智慧融入AI系 ...

智通财经7月14日电，澳大利亚悉尼大学研究团队开发出一种名为PROTEUS的"生物人工智能（AI）"系 统。该系统通过模仿自然进化过程，能在几周内创造出具有新功能的分子，为药物研发和生物技术带来 突破性进展。该成果发表在最新一期《自然·通讯》期刊上，不仅展示了生物技术与AI融合的前景，更 可能为个性化医疗和精准治疗提供关键工具。 "生物人工智能"系统创建 ...

Core Viewpoint - Black Sesame Intelligence International Holdings Limited (stock code: 2533.HK) plans to acquire an AI chip company through equity acquisition and capital injection, focusing on high-performance, low-power AI system-on-chip (SoC) solutions [1] Group 1 - The target company specializes in the development and sales of AI SoC solutions, with a significant portion of its intellectual property (IP) being self-developed, including image signal processors (ISP), neural network processors (NPU), and analog IP [1] - The acquisition aims to expand Black Sesame's product line across high, medium, and low-end computing chips, enhancing its capabilities in providing comprehensive solutions for smart vehicles and robots [1] - Black Sesame has established partnerships with several automotive companies, such as Geely, Dongfeng, and BYD, and is accelerating its expansion into the robotics chip market [1] Group 2 - The target company is expected to meet specific criteria, including a high self-research rate of core IP, achieving high-performance AI inference within a power range of 20W-50W, and obtaining AEC-Q100 Grade 1 certification [2] - It should occupy a technological advantage in areas like autonomous driving perception and have technical reserves for robotics and industrial automation applications [2] - The company should have experience in advanced process manufacturing below 7nm and stable foundry resources, with its toolchain and algorithm library compatible with Black Sesame's existing platform [2]