Skip to content

103. Smart Replay modes

There are two primary modes to control how vehicles follow the original trajectory: exact mode and behavioral mode. The system can transition between these modes to balance precise path adherence with dynamic, responsive actions.

103.1 Exact Mode Replay

In exact mode, all vehicles replay the exact trajectory from the original run. This mode enforces strict trajectory adherence by feeding the recorded trajectory directly into the vehicle's controls. For runs derived from real-world data, this mode utilizes the original run's time steps.

This mode is ideal for scenarios where maintaining a strict, repeatable trajectory is critical, such as for performance evaluations or controlled tests.

103.2 Behavioral mode replay

In behavioral mode, vehicles aim to follow their original recorded trajectories, but are free to deviate in response to their environment. This mode features:

  • Dynamic adaptation: A vehicle can adjust its path to respond to changing conditions, such as the actions of surrounding traffic or to avoid potential collisions.

  • Intelligent path tracking: The system continuously calculates an expected time and position along the original path. If the vehicle deviates, its objectives are updated, and it will attempt to return to the original path when it is safe and practical to do so.

This mode allows for more realistic and adaptive vehicle behavior in complex, interactive scenarios.

103.3 Transitioning from Exact to Behavioral mode

The transition mechanism is designed to maintain alignment with the original run and change to behavioral mode only when the scenario necessitates it.

103.3.1 During a standard run

When an actor is configured to use behavioral mode, the transition to behavioral mode happens only when the initial warm-up period has finished.

103.3.2 When applying scenario modifications

When real-time modifications are applied to actors, the system switches the affected vehicles to behavioral mode after the warm-up period.