3 6 Ke·2025-10-20 08:33Core Insights - The automatic driving industry is experiencing a battle for narrative control over next-generation technologies, with companies like Li Auto and XPeng betting on VLA (Visual Language Action) as the future architecture, while Huawei criticizes it as a shortcut and promotes its own WA (World Behavior Architecture) [1][2][3] - The rapid emergence of jargon in the industry reflects the struggle for technological branding, as hardware becomes increasingly homogeneous and intelligent driving capabilities become the key differentiator [1][2][3] Group 1: Evolution of Terminology - Before 2022, the automatic driving industry's technical evolution was primarily defined by Tesla and Waymo, with terms being objective descriptions of specific functions [3] - Tesla's AI Day events in 2021 and 2022 significantly influenced the industry, introducing the BEV+Transformer architecture, which improved perception capabilities by integrating multiple camera inputs into a unified 3D view [3][4] - The transition to an "end-to-end" paradigm began in 2022, breaking down the barriers between perception and planning, with Tesla's FSD Beta V12 showcasing a large neural network that processes both simultaneously [5][6] Group 2: Technological Developments - Chinese automakers quickly adopted Tesla's advancements, with companies like XPeng and NIO implementing their own versions of the BEV+Transformer architecture for mass production [4][6] - The industry is moving towards a more integrated approach, with XPeng and Huawei adopting multi-stage end-to-end systems, while NIO is restructuring to focus on end-to-end development [7][8] - The introduction of VLA and world models into the automatic driving sector reflects a shift towards more sophisticated AI models that can understand and respond to complex driving scenarios [9][10][13] Group 3: Competitive Landscape - The competition in computing power is intensifying, with XPeng and Li Auto investing heavily in both vehicle and cloud computing capabilities, aiming to develop larger parameter models for their systems [11][12][36] - The VLA model, initially developed for robotics, is being adapted for automatic driving, with companies like Yuanrong Qixing leading the charge in applying this technology [10][31] - NIO and Huawei are taking a more aggressive approach by deploying world models directly in vehicles for real-time control, although the technology is still in the experimental stage [14][15] Group 4: Future Directions - The evolution of automatic driving terminology indicates a broader exploration of technology, with each new term representing a step in the industry's journey [16] - The ultimate success in the automatic driving sector may hinge on the ability to translate technological promises into tangible user experiences, rather than merely introducing new concepts [16]