GM to launch highway self-driving in 2028, expand to SUVs and pickups

[Financial News] U.S. automaker General Motors (GM) plans to introduce highway self-driving technology in 2028 and extend it to its core sport utility vehicles (SUVs) and pickup trucks. The strategy is to accelerate the mainstream adoption of self-driving technology by building a next-generation computing system that can be used in both electric and internal-combustion vehicles. GM is also moving to center vehicle development on artificial intelligence (AI) and software as it seeks to secure a leading position in the future mobility market.
GM said on the 23rd that it had unveiled next-generation self-driving technology and the electric and computing architecture that will support it at the GM Forward event, held in New York on the 21st (local time).
GM plans to introduce highway self-driving on the Cadillac Escalade IQ, a large electric SUV, starting in 2028. It is currently testing the technology on a closed course using pre-production vehicles and plans to extend it to key internal-combustion SUVs and pickup trucks in the future.
At the core is GM’s next-generation central computing unit (CCU), developed in-house. The liquid-cooled CCU delivers about 40 times the computing performance of existing vehicles. It is designed to integrate software for self-driving, active safety, the chassis and propulsion into a single high-speed system, enabling in-vehicle AI functions and the self-driving system to run simultaneously.
Another feature of the next-generation architecture is that it separates hardware from software. The same computing, network and software systems will be used in electric and internal-combustion vehicles, while control functions will be tailored to the characteristics of each vehicle model. This is expected to reduce the cost and time required to develop and implement new technologies.
The vehicle’s internal communication network will also be significantly upgraded. Zone controllers connected via high-speed Ethernet will be able to process 36 times as much data as before. Wiring length and weight will be reduced by 30%, while cellular connection speeds will increase by up to 50%.
GM plans to first apply the next-generation architecture to the Escalade IQ in 2028, then extend it to key full-size internal-combustion SUVs and trucks, three-row crossovers and fourth-generation electric vehicles. It will also establish a system for applying improvements validated on one model to other models through over-the-air (OTA) software updates.
GM is also stepping up efforts to amass large-scale data and train AI to improve the performance of its self-driving technology. The cumulative hands-free driving distance logged by customers of GM’s Super Cruise driver-assistance system has surpassed 1.2 billion miles (about 1.93 billion kilometers).
GM uses driving data from customer vehicles whose owners have consented to data collection, while also deploying hundreds of dedicated vehicles to gather about 1 million miles of multisensor data each month. The information collected on real roads is used to train AI and improve self-driving systems.
GM is also expanding testing in virtual environments. By combining reinforcement learning and large-scale simulations with imitation learning based on real-world driving data, it is improving the system’s ability to handle a range of unexpected situations. In particular, it simulates at least 3 billion miles of driving a day, repeatedly training on situations that are unlikely to occur on real roads.
GM aims to improve the safety and reliability of its self-driving technology through this data-driven development system and expand its application, focusing on high-volume SUVs and pickup trucks.
[email protected] Kim Dong-ho Reporter