Computer Vision in Vehicle Technology

Chapter 2Autonomous Driving

Uwe Franke

Image Understanding Group, Daimler AG, Sindelfingen, Germany

During the last two decades, unprecedented progress has been achieved in computer vision research. At the same time, the quality of images delivered by modern CMOS imagers in outdoor scenarios has improved significantly and the available computational power constantly doubled in less than 2 years. As a consequence, computer vision for cars has evolved from simple lane keeping to powerful driver assistance systems, as described in Chapter 7. The most ambitious and highly attractive research goal is Autonomous Driving, especially through busy urban traffic.

This chapter is organized as follows: first, we give some general remarks on Autonomous Driving and the environment perception for autonomous vehicles. Subsequently, we concentrate on image understanding for complex urban traffic scenes. Finally, ongoing research topics are outlined.

2.1 Introduction

2.1.1 The Dream

The dream of Autonomous Driving is nearly as old as the car itself. As early as 1925, Francis P. Houdina publicly demonstrated his radio-controlled driverless car “Linrrican Wonder” on the streets of New York City, traveling up Broadway and down Fifth Avenue, through the thick of the traffic. Some years later at the 1939 New York's World Fair, Norman Bel Geddes designed the Futurama exhibition that presented a possible model of future traffic including automated highways where cars would drive on inductive cables ...