IntersectionBehaviorΒΆ
Depending on the Input of the Sensor_Junction and the assessment in the previous modules, the agent derives an action depending on (relevant) signs and (relevant) agents. Due to the modular approach, it is possible to only use the module ActionDeduction with the Sensor_Junction which would lead to a reaction on every surrounding agent and traffic sign. The action of the agent is also represented by a velocity wish for the next timestep. If this is smaller than the velocity wish from the previous described methods, than it is passed on.
For each relevant traffic sign a reaction is derived. Currently only the stop on a stop-sign is implemented. For this, the agent stops on the sign for 3s and then continues its ride depending on the relevant surrounding agents. For each relevant agent a reaction is derived, which is usually a target-brake on the junction entrance. The agents where already assessed through the module SituationAssessment, so only agents with a possible higher rank, collision path or right of way are passed. If the agent can pass the conflict zone before the relevant agent enters it, the velocity stays. If there is a critical intersection (ttc < 3s), then a emergency break is requested.
Different driving directions currently need to be separated by a solid line because the current implementation in der sensor-driver only detects the absolute yaw-angle for other objects and the ownVehicleInformation contains only the relative yaw angle. No information about the lane directions are passed so the lane-change-model isnt able to react on a oncoming vehicles properly.