Enhanced Traffic Simulation for Improved Realism in Driving Simulators

Bihao Wang, Marwa Mahmoud, Hannah Close, Lee Skrypchuk, Javier Echevaria Cuesta, Quentin Stafford-Fraser, Peter Robinson
ACM Conference on Automotive User Interfaces, Toronto, Canada, September 2018.
The ability to monitor and detect potentially dangerous behaviour in surrounding traffic is vital for the development of intelligent vehicles. However, data collection for these kinds of scenarios is difficult in real-life, and a driving simulator therefore becomes an important substitute. In this paper we present an approach to enhance driving simulators. We experiment on an open source development platform, which is used to test real-life use cases within a simulated vehi- cle environment. We propose replacing pre-programmed traffic dynamics with real driving data recorded from human drivers in the same environment. This enhances the engagement of the host driver in the more realistically simulated traffic scenario. Signal lights and indicator sounds are also integrated to enrich the driver’s sensation. Our preliminary quantitative and qualitative evaluation shows that our enhanced traffic simulation results in an improvement to the driver’s perception of the realism of the driving simulator.

A short clip showing part of the system in use:

Available here: PDF