Driver Reaction Model Structure - JnSmmr/OpenPASS-Wiki GitHub Wiki

The Driver Reaction Model is built from several components that work together to best simulate the behaviour and specific reactions of a human driver inside the simulation framework. The following should give a good overview about the different parts of the module as well as how they interact.

DReaM Modules

(TODO: add updated/finished UML Diagram & shortly discribe it.)

Interpreters

Collision Interpreter

Crossing Info Interpreter

Following Interpreter

Right of Way Interpreter

World Interpreter

Gaze Movement

Action Decision

Navigation

Agents use their navigation module to determine their path in the simulated environment. Therefore it uses information from the WorldRepresentation as well as the Route and Waypoints imported from the scenario.xosc file. With every timestep, it is checked wether or not the current navigation goal is already set. If not, the shortest path to the next waypoint is calculated and an updated navigation decision is provided.

CognitiveMap

TrafficSignMemory

A simplified walk-through through the simulation

Simulation start up

• import configuration files and convert the information
• build a representation of the 'world' described by them
• spawn agents according to the scenario's given rules

What happens every timestep?

For reasons of clarity, a rough overview throughout a single agent's actions are considered.

• sensor converts and integrates information into an internal road network structure
• sensor provides information visible/tetectable for the agent
• information is forwarded to the main components of DReaM
• environment informations are interpreted and processed
• calculated and applied to the agent are: TODO: korrekte namen!
  • target acceleration
  • lateraler versatz
  • differenzwinkel
  • turning indicator state
  • break status

Simulation finish

• Information stored in the AgentStateRecorder is written to DReaMOutput.xml