General Motors Starts Real-World Testing of Advanced Automated Driving Systems
General Motors has entered a pivotal new phase in its journey toward fully automated driving, marking a significant step forward in the development and deployment of next-generation autonomous vehicle technologies. The company recently announced the start of supervised public-road testing for its latest automated driving system, a move that reflects years of research, large-scale data collection, and extensive simulation-based validation.
Transitioning from Data Collection to Real-World Testing
For years, GM has been quietly building the foundation for its next-generation automated driving platform. This effort has included collecting and analyzing massive amounts of driving data across diverse road conditions, geographies, and traffic environments. With over one million miles logged by GM’s data collection vehicles across 34 U.S. states, the company has accumulated a robust dataset that now underpins its newest system.
The shift to supervised public-road testing represents a critical transition from passive data gathering to active system validation. Instead of simply observing human driving behavior, GM’s vehicles are now engaging directly with live traffic environments on limited-access highways in states such as California and Michigan. These highways provide a controlled yet realistic setting, allowing engineers to evaluate system performance under consistent speeds, structured lane usage, and predictable traffic patterns.
This phase is not fully autonomous. Each test vehicle is equipped with a trained safety driver seated behind the wheel, ready to take control at any moment if needed. This supervised approach ensures that safety remains the top priority while enabling the system to learn and adapt in real-world scenarios.
Scaling Up the Test Fleet
In the coming months, GM plans to expand its testing program significantly. More than 200 development vehicles—both manually operated and supervised automated units—will be deployed on public roads. This large-scale rollout will allow the company to gather a broader range of real-world data, covering different driving behaviors, weather conditions, and regional traffic dynamics.
By operating within live traffic environments, these vehicles will encounter unpredictable situations that cannot always be replicated in simulations. These include sudden lane changes by other drivers, varying road conditions, construction zones, and unexpected obstacles. Such exposure is essential for refining the system’s decision-making capabilities and ensuring it can handle the complexities of everyday driving.
A Disciplined Path Toward Autonomy
GM’s approach to automated driving development has been methodical and incremental. Rather than rushing to deploy fully autonomous systems, the company has emphasized a phased strategy that balances innovation with safety and reliability. Each stage of development builds upon the previous one, ensuring that new capabilities are thoroughly tested before being introduced to consumers.
This disciplined methodology is evident in the progression from data collection to supervised testing, and eventually to hands-free and eyes-off driving systems. By validating each layer of technology in controlled conditions, GM aims to minimize risks and build trust among consumers and regulators alike.
Introducing Eyes-Off Driving by 2028
A major milestone in GM’s roadmap is the planned introduction of “eyes-off” driving technology by 2028. This advanced system will allow drivers to disengage from continuous monitoring of the road under specific conditions, representing a significant leap beyond current driver-assistance systems.
The first vehicle expected to feature this capability is the Cadillac Escalade IQ, a premium electric SUV that will serve as the launch platform for GM’s next-generation autonomy. Following its debut, the technology will be gradually expanded to other vehicles across GM’s portfolio, including both electric and internal combustion models.
Initially, eyes-off functionality will be limited to highway driving, where conditions are more predictable and easier to manage. Over time, GM plans to extend this capability to more complex scenarios, eventually enabling “driveway-to-driveway” automation that can handle an entire journey from start to finish.
A New Centralized Computing Architecture
One of the key enablers of GM’s automated driving ambitions is its new centralized computing architecture. Unlike traditional vehicle systems that rely on numerous distributed control modules, this architecture consolidates processing power into a unified platform.
This approach offers several advantages. It allows for faster data processing, more efficient software updates, and greater scalability across different vehicle models. Importantly, it enables GM to deploy advanced features like eyes-off driving across a wide range of vehicles—from high-end luxury models to more affordable mainstream options—without needing to redesign the system for each platform.
The centralized architecture also supports over-the-air updates, ensuring that vehicles can continuously improve over time as new software enhancements are developed and deployed.
Leveraging Real-World and Autonomous Driving Data
A significant strength of GM’s automated driving program lies in its access to vast amounts of real-world driving data. The company’s Super Cruise system, one of the industry’s leading hands-free driver assistance technologies, has already accumulated more than 800 million miles of customer-driven data across 23 vehicle models.
This dataset provides invaluable insights into real-world driving behavior, helping engineers refine algorithms and improve system performance. By analyzing how drivers interact with the system, GM can identify areas for improvement and enhance the overall user experience.
In addition to Super Cruise data, GM benefits from the contributions of Cruise LLC, its autonomous vehicle subsidiary. Cruise has logged more than 5 million miles of fully autonomous driving without a human driver, primarily in complex urban environments. These experiences provide a complementary perspective, exposing the system to challenging scenarios such as dense traffic, pedestrian interactions, and intricate road networks.
By combining these two sources of data—customer-driven and fully autonomous—GM is able to develop a more comprehensive and robust automated driving system.
The Role of Simulation in Accelerating Development
While real-world testing is essential, it is not sufficient on its own. To accelerate development and ensure comprehensive validation, GM relies heavily on advanced simulation technologies. The company’s simulation environment is capable of replicating approximately 100 years of human driving behavior every single day.
This immense computational capability allows engineers to test the system under a wide range of scenarios, including rare and dangerous situations that would be difficult or unsafe to encounter in real life. These simulations help identify potential edge cases, refine system responses, and improve overall reliability.
By integrating simulation with real-world data, GM can iterate quickly and efficiently, continuously enhancing the performance of its automated driving system.
Building Trust Through Transparency and Innovation
As automated driving technology continues to evolve, building public trust remains a critical challenge. GM recognizes that transparency and communication are key to addressing concerns and fostering confidence among consumers.
To this end, the company has launched the GM Engineering blog, a platform designed to share insights into its technological innovations and development processes. Through this initiative, GM aims to provide a deeper understanding of how its systems work and the steps being taken to ensure safety and reliability.
By openly discussing its progress and challenges, GM hopes to demonstrate its commitment to responsible innovation and establish itself as a leader in the field of intelligent vehicles.
Source Link:https://news.gm.com/
