大晓机器人发布“以人为中心”ACE具身研发范式

Core Insights - The company, Daxiao Robotics, has introduced the ACE embodied research paradigm and the open-source Kairos 3.0 model, aiming to create a self-controlled and collaborative industrial ecosystem in the embodied intelligence sector [1][2]. Group 1: Industry Challenges - The embodied intelligence industry faces a significant data bottleneck, with real machine data limited to 100,000 hours, creating a substantial gap compared to the vast data available in the autonomous driving sector [1]. - Traditional research paths relying on real machine remote control or pure visual learning encounter challenges in cost, efficiency, and generality due to the "machine-centric" development paradigm [1]. Group 2: Technological Innovations - The ACE paradigm emphasizes a "human-centric" approach, shifting the focus from hardware parameters to human interaction in the development process [1]. - Environmental data collection technology, a core component of the ACE paradigm, integrates multi-dimensional data from various modalities, enhancing the training of embodied intelligence models [3]. - This technology enables comprehensive data coverage, supporting complex tasks with hundreds of atomic actions and achieving sub-pixel level precision in tracking intricate human movements [3]. Group 3: Collaborative Ecosystem - Daxiao Robotics has partnered with various hardware manufacturers, including Insta360 and Wolong Electric Drive, to enhance the information collection capabilities of their models and modules [4]. - In the cloud services and data domain, collaborations with companies like SenseTime and Tencent Cloud aim to build a comprehensive support system, reducing R&D costs for small and medium-sized enterprises [4]. - The Kairos 3.0 model has been adapted to work with several domestic chip manufacturers, significantly improving chip performance and fostering a collaborative synergy from foundational computing power to intelligent applications [2][4].