External simulation model that can be linked to SUMO to enable realistic left turning behavior of cyclists at a single signaled intersection. To enable realisitc bicycle behavior at multiple intersections, multiple intersection_model instances need to be run. In that case, the given implementation needs to be forked so that the intersection_model does not start the SUMO simulation when executed, but is linked to a running SUMO simulation.
The model was implemented and tested using Python3.6.
To run the model, the respective variables in settings.py need to be set. Then run main.py.
By default, the model runs an exemplary scenario featuring the intersection Alexanderstr./Karl-Marx-Str. in Berlin.
Note that when evaluating the model's hyperparameters (settings.eval_hyperparameters=True), the sumoBinary should be set to path/to/sumo/bin/sumo to speed up the evaluation. Furthermore, it was found that the model performs sufficiently faster when being started from outside the IDE.