The project utilizes the Carla simulator and a racing wheel to create a platform that aimes to check the hypothesis that drivers unconsciously swerve toward roadside obstacles due to diverted attention.
Carla is an open-source Simulator developed to support a variety of aspects of autonomous driving research.
The simulator can be used to recreate real world traffic events and to collect tagged data from these events for training autonomous driving systems.
In one of the later versions of the Carla simulator, an option to take over control of an object in the simulated world was added.
This object that can be controlled by a racing simulator steering wheel and pedals and interacts with the simulated world and with other objects in it, Like in a video game.
Cyclists using the road shoulder and other highway users that stop by the road are often hit by on-going traffic on the highway. Due to the large differences in speed, these accidents are extremely deadly.
Thus, studying and understanding their causes is important to reduce their occurrence.
Traffic research has shown that drivers don’t tend to drive off into the shoulder. However, in instances where the shoulder is occupied by a cyclist or a stopped vehicle, the odds of a highway user driving by swerving into the shoulder and causing a collision are higher.
Due to this fact, it is hypothesized that drivers that notice another road user on the shoulder tend to swerve in their direction unknowingly and subconsciously,
as a direct result of the drivers’ attention being distracted by the road user on the shoulder.
In this project this hypothesis is tested by creating a scene that persists of a driver overtaking a cyclist that is on the shoulder using the Carla simulator and accessory simulator equipment like a racing steering wheel and pedals to best recreate a real-world driving experience.
Also, the compatibility of Carla’s API and accessory simulator equipment with this type of experiment is tested for future use in the laboratory.
