Lane Localization for Autonomous Model Cars

Lukas Maischak— 2014

Mobile robotics is a rapidly growing field and has countless applications including exploration, logistics, rescue operations, as well as domestic and military use. One particularly interesting example of ist use is the construction of autonomous, "self-driving" cars. Imagine that car accidents caused by human error are a thing of the past, or that your car can find ist own parking spot after you have left the vehicle. In many cases, mobile robots need to plan and make decisions autonomously while interacting with their environment. A necessary prerequisite for them to execute most non-trivial tasks is to have a concept of their environment and their location in it. Determining this location is a fundamental problem in mobile robotics known as localization. Autonomous cars need to know where they are on the road, both on a small scale to stay in lane and on a large scale to navigate. This project explores mobile robotic problems on a small scale. The objective is to construct a small model car that is able to drive autonomously in an indoor model environment like those displayed in Figures 1.1 and 1.2. This document focuses on a mechanism for localizing autonomous model cars within such an environment. The next chapter gives a brief overview of the target platform - the model car in use. Chapter 3 defines the problem this document is triying to solve in more detail. Chapters 4 and 5 describe and evaluate two alternative - though related - solutions to that Problem.