In English

Synthesising Data from Subsystems in an Active Safety System - Design of a Data Processing Tool

Alvean Ekman ; Daniel Axman
Göteborg : Chalmers tekniska högskola, 2015. 60 s.
[Examensarbete på avancerad nivå]

This project proposes a design of a data processing tool for a Trac Sign Recognition(TSR) system, as a part of an Active Safety System that utilizes a front-mounted radarcamera set-up. Data from other subsystems that contribute to completing processed data from a utility perspective is aggregated as well. The tool is to be used for synthesising large amounts of data from log les generated during post-analysis of test drives, thereby increasing eciency of testing and veri cation of the system.

A Python script functions as a connector between MATLAB structures storing information recorded during the test drives and a storage database. Data of interest is extracted from the MATLAB structures generated in post-analysis of test drives, then handled by a Python script that opens a connection to a PostgreSQL database in which the data is inserted. Conclusions concerning the accuracy of the TSR system can be drawn from evaluating the data in the database.

This thesis contributes to improvement of the current testing methods of the above mentioned systems, and hence utilises some existing procedures and functions at Delphi Automotive. The tool uses previously recorded and classi ed data for comparison, and is not intended for stand-alone use. Processing of data is not performed real-time in the vehicle, but during analysis of the data afterwards. The prototype tool is developed in connection to Swedish trac. The test drives are performed under applied conditions, the input to the system is not simulated. The purpose of this thesis is to design a prototype tool to ful l the above requirements, a complete tool is thus not implemented.

Furthermore, the systems handling extraction and analysis of data in the vehicle are treated as black boxes. There are protected material which cannot be accessed for the purpose of a thesis, and thus the information obtained from these systems cannot be dissected as to how they came to these conclusions.

The conclusion of this project is that it is advisable to continue development of such a tool as it can improve existing testing and veri cation procedures. There are a number of areas that need to be addressed, for which recommendations are given, none of which are deemed impossible to solve. This thesis suggests a design of a data processing a tool and a number of functionalities that can be implemented in a future project.

Publikationen registrerades 2015-03-23.

CPL ID: 214177

Detta är en tjänst från Chalmers bibliotek