Thesis

Laminar traffic flow: a continuous flow algorithm for intersecting traffic of automated vehicles

This Computer Science Masters Thesis sets out to explore an algorithmic approach to traffic regulation of automated vehicles in a single intersection. For this purpose, an X-shaped intersection with randomized oncoming traffic is simulated using Unity 3D in C#. The Laminar Traffic Flow algorithm (LTF), written for the purposes of this thesis, is then used to modulate each arriving vehicle's speed in a way that would allow it - regardless of its size, speed, or turning plan - to drive through the intersection without stopping or colliding with other vehicles. Analysis of LTF's performance when compared to a simulation of a standard, traffic-lights-regulated intersection, shows LTF to have the following advantages: Higher throughput: Up to 20% more vehicles per second go through the intersection. Lower through-time: Vehicles going through the intersection do not need to stop, and maintain their average speed to give minimal through-time, as if the intersection is empty. The LTF algorithm did however demonstrate the disadvantage of having a failure point, a certain amount of incoming traffic load past which it cannot find a solution. How this failure point can be avoided is discussed in the conclusion (Chapter 7). With recent and future advances in traffic automation, the LTF algorithm may be of use in many areas where traffic needs to be regulated, such as intersections, railway intersections, robotic warehouses, and drone control.

Relationships

Items