*This content was translated by AI.View Original
General Motors (GM) officially unveiled its roadmap for expanding autonomous driving across key sales models, along with its next-generation electric and computing architecture, during the "GM Forward" event held in New York, USA, on October 21.
GM plans to introduce highway autonomous driving functionality to the Cadillac Escalade IQ starting in 2028. Currently, testing is underway on closed courses using pre-production vehicles from Yangsan, with the scope of application set to expand to major internal combustion institutional investor models in the future. Following its application to the Cadillac Escalade IQ in 2028, this architecture is scheduled to be sequentially expanded to major internal combustion institutional investor full-size SUVs, trucks, three-row crossovers, and the fourth-generation EV LINEup.
The next-generation architecture separates hardware from software and establishes a computing, network, software, and interface system that can be commonly applied to both internal combustion institutional investor vehicles and electric vehicles. It is designed to apply the same technological foundation across various models while maintaining the hardware and control functions required for each vehicle's purpose. The in-house designed liquid-cooled Central Computing Unit (CCU) provides computing performance approximately 40 times higher than current standards, integrating autonomous driving, active safety, chassis, and drive software. The high-speed Ethernet-based zonal aggregation device processes 36 times more data than existing systems, reduces wiring length and weight by 30%, and improves cellular connection speeds by up to 50%.
Through the Software-Defined Vehicle (SDV) structure, improvements validated in specific models are applied across the entire LINEup via OTA updates. Autonomous driving development and verification utilize 1.2 billion miles of hands-free driving data accumulated from SuperCruise, along with approximately 1 million miles of multi-sensor data collected monthly by hundreds of dedicated vehicles. Additionally, it enhances responsiveness by combining imitation learning, self-play-based reinforcement learning, and large-scale simulations that virtually drive over 3 billion miles daily.
<© STARNEWS. All rights reserved. No reproduction or redistribution allowed.>
*This content was translated by AI.View Original
