I. INTRODUCTION
The interest in autonomous cars has grown in recent years,
as they provided new insights for general robotic systems
in areas like safety, machine learning and environmental
perception [2].
One key aspect for driving autonomous cars is the detection
of obstacles and other cars. In our paper we focus on a
highway scenario. Here, we describe an algorithm which
enables our car to follow other cars at various speeds, while
keeping a safe distance and providing braking in front of
obstacles. We describe how data from Lidar and radar can
be used and combined for precise obstacle and car detection
at different velocities.
Sensor fusion of Lidar and radar combining the advantages
of both sensor types has been used earlier, e.g., by Yamauchi
[14] to make their system robust against adverse weather
conditions. Blanc et al. [1] present an application that focuses
on the reliable association of detected obstacles to lanes and
the reduction of false alarms (phantom obstacles). A variety
of applications of sensor fusion methods in cars are given by
Kaempchen et al. in [7].
The remainder of this paper is structured as follows: Section
II presents our data fusion approach. Section III and
Section IV illustrate how the obstacles are incorporated in
the path-planning process and velocity calculation for the
autonomous car. Experimental results during a test run on
public highways are given in Section V. Finally, Section VI
draws conclusions and discusses future work.