An approach to simulate autonomous vehicles in urban traffic scenarios

Abstract

The most common cause of traffic accidents is arguably the driver error due to lack of attention. And it is very unlikely this is going to change soon thanks to increasingly cell-phone usage, in-car entertainment systems, and naturally the more frequent traffic jams in highly populated areas. Autonomous vehicles, such as driverless cars, are a promising approach to decrease traffic accidents, as well as congestions. To test this approach, simulations are a safer, more efficient, and cheaper way than live testing. This paper presents an approach to implement a simulator to test such vehicles. It includes a study of the state of the art in driverless car simulation and discusses on the specific objectives that this particular simulator aims to achieve in order to aid testing the interactions of multiple driverless cars in urban networks.