GM supervised highway testing for next generation autonomous driving system

Expanding its real-world validation strategy, General Motors has initiated supervised public-road trials of its advanced driver automation platform across selected highways in California and Michigan. The initiative represents a critical step toward deploying its next-generation system, with over 200 development vehicles now operating in live traffic scenarios. Each vehicle is equipped with trained safety drivers to ensure immediate manual intervention when required, enabling safe yet data-rich testing conditions.

Supervised Highway Testing Across Key U.S. Regions

The latest phase focuses on limited-access highways, where controlled environments allow more predictable traffic patterns while still providing real-world complexity. By deploying a large fleet of manually supervised vehicles, GM aims to validate system performance under diverse driving conditions. These vehicles are continuously monitored, ensuring that the system’s behavior aligns with safety expectations while gathering high-quality operational data.

Real-World Data Collection and Scale

The company has already accumulated over one million miles of driving data across 34 states, highlighting the scale of its development program. This new testing phase significantly enhances the depth and diversity of collected data. Each mile driven contributes to refining perception algorithms, decision-making frameworks, and system reliability under varying environmental and traffic conditions.

AI Model Enhancement Through Continuous Feedback

All operational data captured during supervised testing is automatically fed back into GM’s development ecosystem. This closed-loop system enables rapid iteration of AI models, allowing engineers to identify edge cases and improve system robustness. The feedback mechanism ensures that learnings from real-world scenarios directly influence software updates, strengthening overall system performance without requiring extensive redesign cycles.

Improving System Robustness and Safety

The integration of real-time data with simulation and validation tools helps GM address complex driving scenarios more effectively. This approach ensures that the system evolves continuously, improving reliability before broader deployment. The presence of trained drivers further enhances safety during this phase, ensuring that unexpected situations are managed without risk.

Roadmap Toward Eyes-Off Driving Capability

In October 2025, GM confirmed plans to introduce an eyes-off automated driving system by 2028, beginning with the Escalade IQ. Unlike supervised systems, this capability will not require continuous driver monitoring for safety. The rollout will initially target highway environments before expanding to more complex routes, including driveway-to-driveway functionality.

Scalable Deployment Across Vehicle Portfolio

One of the key differentiators in GM’s strategy is scalability. The automated driving system is being designed to function across multiple vehicle segments without requiring a complete redesign for each model. This allows deployment from premium Cadillac vehicles to mainstream Chevrolet offerings, ensuring broader accessibility while maintaining consistent performance standards across the portfolio.

The ongoing supervised testing phase plays a foundational role in achieving this scalability, ensuring that the system is adaptable, robust, and ready for mass-market integration in the coming years.