Traffic State Estimation

The traffic state estimation is a crosscutting tasks used by multiple layers. The main goal is to transform the surrounding traffic and lane configuration into the vehicle's Frénet Frame and estimate their future state. That includes all static objects, regulatory element locations, pedestrians, and vehicles (including Ego) along the reference path.

Since the traffic is in constant change, we estimate this state by evaluating the SDV current trajectory and the predicted time until a new Trajectory can be generated by the Maneuver Layer. This predicted traffic snapshot in Frénet frame mitigates the complexity in the decision-making process and trajectory generation with a simplified representation of the world geometry and dynamic agents.