Alexander Kirchner

Integrated obstacle and road tracking using a laser scanner

Autonomous vehicles need sensory input from the environment. Here, a laser scanner mounted on the front bumper is used to provide range data of a horizontal profile in front of the vehicle up to a maximum distance of one hundred meters. Other vehicles are detected and tracked in successive scans yielding their position and velocity vectors. The road boundary consisting of crash barriers and reflector poles is used to estimate the parameters of a clothoid model. The combination of object tracking and boundary estimation results in both a more robust and accurate estimation and enhanced knowledge of the overall traffic situation. By taking a priori assumptions into account a lane marking of vehicles is possible. Results are shown with recorded real traffic data.

Optimization of Sensor, Bus, and Fusion Schedules of a Time-Triggered Sensor Fusion System

In this paper, we optimize a time-triggered, Kalman filter based, multi-sensor fusion system, used as an environmental perception platform for advanced driver assistance systems while satisfying constraints that are typical of a safety-related application. We argue that the overall system including effects from the sensor, bus, and fusion schedules as well as the treatment of measurements must be considered in order to optimize the fusion accuracy. Due to differences in measurement preprocessing data from sensors may not arrive in chronological order which requires special treatment for this out-of-sequence measurements (OOSM). As a result of this paper we identify regions in the scheduling parameter space that minimize the error covariance of the estimated states.

Ein fortgeschrittenes Kollisionsvermeidungssystem

Fahrerassistenzsysteme haben wesentlich zur Erhohung der Fahrzeugsicherheit beigetragen. Langfristiges Ziel dieser Entwicklungen stellen Kollisionsvermeidungssysteme dar. In Zusammenarbeit zwischen der Volkswagen AG und der Helmut-Schmidt-Universitat/Universitat der Bundeswehr Hamburg entstand im Rahmen des Invent-Projektes ein System, das autonom Brems- oder Ausweichmanover ausfuhren kann.

Analysis of sensor and fusion schedules of a time-triggered sensor fusion system

In this paper we analyze the sensor and fusion schedules of a time-triggered, Kalman filter based, multi-sensor fusion system. The fusion system is used as an environmental perception platform for advanced driver assistance systems and delivers its service to a safety related application. As the application demands cyclic updates with bounded accuracy, the influence of the sensor and fusion schedules on the service accuracy is analysed, which enables us to optimize the system without model dependent simulations.

The electronic copilot for an autonomous vehicle-state of development

Gives an outline of the current state of development of an electronic copilot (ECP). The ECP is used within an autonomous vehicle as a partially redundant system for the vehicle control. The ECP receives the necessary information about the positions and relative velocities of obstacles from various sources such as laser scanner or vision sensors. The algorithms for emergency braking and avoiding stationary obstacles are presented. If the time-to-collision falls below the braking- or evading-time the ECP starts to intervene. Experimental results show the capabilities of the system.

An advanced collision avoidance system

Driver assistance systems have made a significant contribution towards improving vehicle safety. The long-term objective of such developments is to produce a complete collision avoidance system. Within the framework of the Invent project and as part of a collaboration between Volkswagen AG and the Helmut Schmidt University/University of the Federal Armed Forces in Hamburg, a system has been developed that can perform braking and avoidance manoeuvres autonomously.