火山引擎具身智能闭门研讨会：大模型驱动具身智能产业新未来

Group 1 - The seminar hosted by Volcano Engine and NVIDIA focused on the new opportunities in the embodied intelligence industry driven by large models, attracting over 50 executives, technical experts, and investors from various fields [1] - Zhang Xin, Vice President of Volcano Engine, highlighted that 2025 will be the year of practical applications for intelligent agents, with embodied intelligence becoming a focal point for the industry [2] - The spending on generative artificial intelligence is projected to grow over six times in 2024 compared to 2023, with vertical AI fields experiencing a twelvefold increase, indicating an accelerated trend in the large model industry [4] Group 2 - The development of embodied intelligence is transitioning from Robot 1.0 to Robot 3.0, where the integration of large models with robots is expected to comprehensively reconstruct the overall system capabilities of perception, decision-making, and control [4] - Volcano Engine's multi-modal data lake solution addresses challenges in data heterogeneity, high-quality data acquisition, and the lack of professional data infrastructure, establishing a complete data processing and management system [5] - A leading domestic intelligent driving company improved model training speed by 1.5 times and increased GPU resource utilization from 30% to 95% using the LAS Lance + EMR Ray technology architecture, significantly reducing overall costs [7] Group 3 - Zhang Xin emphasized that the development of embodied intelligence requires not only technological innovation but also organizational transformation to foster creativity and agile iteration [8] - Volcano Engine aims to deepen collaboration with ecosystem partners to provide comprehensive technical support for the embodied intelligence industry through large model technology and multi-modal data lake infrastructure [8] - The successful seminar showcased Volcano Engine's technical strength in large models and data infrastructure, creating a platform for deep communication and collaboration within the industry [8]