Home - JnSmmr/OpenPASS-Wiki GitHub Wiki
Welcome to the OpenPASS-DReaM Wiki!
Motivation and Goals
This wiki is a description as well as user guide about the DReaM extension to the OpenPASS framework. It aims to facilitate an understanding of both functionality and usability of DReaM. The Driver Reaction Model - short DReaM, is an extension to the OpenPASS simulation framework for complex traffic situations. As a model working on the basis of OpenPASS simulations, the goal of DReaM is presenting a means of depicting a more realistic view on driver behaviour. This may deem increasingly important due to the rising popularity of driver assistance systems and the automation of driving functions, wich require testing especially towards safety concerns. While the OpenPASS framework presents a basis for running traffic simulations with user-given parameters, DReaM adds the possibility to depict the exact behaviour of the driver more precisely. It is specialized to simulate the more complex urban traffic and excels at analyzing critical situations and accidents. This can be especially helpful for simulating accidents and how they could have been prevented.
Find out more about DReaM at: https://driver-model.de
Contents
While the OpenPASS documentation already gives good insight into the parent project and its setup, the workflow and set of configuration files have to be modified for utilizing DReaM. Henceworth this wiki contains a comprehensive guide through additional requirements and results added to the core framework. Most importantly it contains a detailed description of each configuration and output file exclusive to DReaM and a brief overview of the ones needed by OpenPASS. It also displays the standard work steps from setting the simulation parameters to interpreting the simulation's results and gives an insight into DReaM's modules and internal processes to facilitate an understanding of its functionality.
Workflow and Structure
Configuration Files
Output Files
Version history? Patch notes?
This work has been created by the Christian Siebke research group at the IAD (Institut für Automobiltechnik Dresden)