Ulf Noyer

Generating high precision maps for advanced guidance support

Modern driver assistance systems increasingly support safer and more fuel-efficient driving. To fulfill these tasks information about the vehicle environment is essential. Besides extracting this information from sensors to perceive the environment, it is also possible to use stored static data. These models allow a reliable and fast access to the environmental data without the typical problems faced with the recognition by sensors, like noise or glitches. Nowadays street maps produced for navigation purposes reach a precision of several meters. To support the driver during manoeuvres however a precision one magnitude better is necessary. Positioning technology has steadily improved over the years and will further improve in the near future. Alongside this development it can be expected, that more precise digital maps will gain more importance. In this paper a method is presented to create these maps mainly automatically. The primary goal for this method is to achieve the best precision possible investing a very small effort. Data is recorded using a test vehicle equipped with a navigation system featuring high precision DGPS and inertial sensors. Lateral deviations are compensated by using a image processing lane-detection sensor. Several measurement iterations are made and merged into a digital map of the street using statistical methods. In order to show the suitability of the generated map for all kinds of ADAS, a test track was surveyed and the digital map was used for automatic guidance of another test vehicle. It could be shown that the map generation algorithm is generally able to produce high precision maps even in challenging environments, however ADAS demanding precise lateral and longitudinal position information can only rely on digital maps and GNSS in environments with little or no obstruction to the sky.

JDVE: A Joint Driver-Vehicle-Environment Simulation Platform for the Development and Accelerated Testing of Automotive Assistance and Automation Systems

As virtualization of design methods in general becomes more and more relevant, one of the main goals of the EU FP7 Project ISi-PADAS is the development of a Joint Driver-Vehicle-Environment Simulation Platform (JDVE) which enables the designers of Advanced Driver Assistance Systems (ADAS) to validate their design with driver models as well as with “real” drivers. In order to cover both test cases, it is necessary to have a highly modular software platform able to be connected to various driving simulators, or even real test vehicles, but also capable of running with a virtual driver model on a single desktop PC. As virtual driver models do not need to act in real time it is beneficial to accelerate their timing in order to cover more test cases, e.g. as application of the Response 3 Code of Practice. This paper explains the modular approach of the JDVE and describes the accelerated time feature. Furthermore it briefly sketches some possible use cases for the JDVE.

Semantic technologies for describing measurement data in databases

Exploration and analysis of vast empirical data is a cornerstone of the development and assessment of driver assistance systems. A common challenge is to apply the domain specific knowledge to the (mechanised) data handling, pre-processing and analysis process. Ontologies can describe domain specific knowledge in a structured way that is manageable for both humans and algorithms. This paper outlines an architecture to support an ontology based analysis process for data stored in databases. Build on these concepts and architecture, a prototype that handles semantic data annotations is presented. Finally, the concept is demonstrated in a realistic example. The usage of exchangeable ontologies generally allows the adaption of presented methods for different domains.