AI road model enables scalable automated driving

HERE Technologies has introduced new Navigation on Autopilot (NOA) capabilities designed to support consistent Level 2++ automated driving across international markets. The solution combines navigation, automated driving functions, and infrastructure awareness within a unified, AI-powered road model to improve predictability and system performance.

Unified road model for automated driving
As automated driving systems move toward large-scale deployment, ensuring consistent behavior across diverse road networks and regulatory environments remains a key challenge. HERE addresses this through a single, AI-driven road model that synchronizes human navigation and machine planning.

This shared “ground truth” enables alignment between what drivers perceive and how vehicle systems interpret road conditions, reducing uncertainty and improving trust in automated driving functions. The model is continuously updated, supporting both development and real-world operation within the digital supply chain.

Predictive signal timing for intersection intelligence
A central component of the solution is predictive traffic signal timing, which integrates real-time SPaT (signal phase and timing) data with historical patterns and machine-learned traffic behavior. This approach provides forward-looking insights into signal changes at intersections.

By improving visibility into one of the most variable driving scenarios, the system enables smoother vehicle responses. Anticipated benefits include reduced abrupt braking and acceleration, more stable speed profiles, improved energy efficiency in electric vehicles, and enhanced accuracy of estimated arrival times aligned with traffic flow patterns.

The development of this capability involves collaboration with ecosystem partners, including Baidu Maps, to support data integration and model refinement.

Lane-level intelligence for NOA systems
The platform introduces lane-level guidance optimized for NOA applications. Instead of defining a single fixed trajectory, the system generates a “route carpet” that identifies multiple viable lane options based on real-time road context.

This allows both automated systems and drivers to anticipate maneuvers such as lane changes, merges, and exits in advance. The result is more consistent vehicle behavior, improved situational awareness, and smoother operation across varying road conditions.

Enabling scalable deployment across markets
The solution supports global deployment of automated driving systems by providing a consistent data foundation across regions. By using the same live map data from development through operation, automakers can train, validate, and deploy automated driving functions with reduced regional adaptation.

Continuous updates to the AI-powered road model maintain data accuracy over time, minimizing discrepancies between expected and actual vehicle behavior. This approach reduces system complexity and supports the scalable rollout of NOA and advanced driver assistance systems across international markets.

Through the integration of predictive intelligence, lane-level mapping, and real-time data updates, HERE Technologies contributes to the advancement of automated driving systems within the automotive data ecosystem.

Edited by Maria Brueva, Induportals editor – adapted by AI.

www.here.com