Jomrich, Florian (2020)
Dynamic Maps for Highly Automated Driving - Generation, Distribution and Provision.
Technische Universität Darmstadt
doi: 10.25534/tuprints-00009702
Dissertation, Erstveröffentlichung
Kurzbeschreibung (Abstract)
With an ever-increasing number of vehicles roaming the streets and a general intensification of ongoing daily traffic the current vehicular safety systems are not able to reduce the number of traffic accidents further.
As the majority of severe or deadly traffic accidents nowadays is caused by human error, car manufacturers and researchers alike focus on the self-driving vehicle as a promising solution to this problem, as a machine is unaffected from human conditions such as tiredness or drunk driving.
To enhance the overall achievable driving safety and comfort the self-driving vehicles rely on an additional map database, besides the hardware sensor system installed onboard.
The so-called High Definition Map (HD Map), a highly precise virtual model of the actual real-world provides detailed information about the ongoing traffic situation ahead of the car's sensor ranges. Otherwise critical traffic situations can be resolved by this a priori knowledge and if necessary, a handover of the driving control back to a human driver can be triggered.
The maintenance of the HD Map is a major challenge, as due to the importance of the map for the self-driving vehicle map updates have to be realized in much shorter time (minutes instead of months) compared to established concepts common for human-oriented digital navigation maps.
This thesis provides contributions in the areas of Distribution, Generation and Provision of such map updates, as the key communication challenges of the maintenance procedure.
Our first contribution is the development, implementation and evaluation of a protocol that realizes the context-specific distribution of partial and incremental map updates. The protocol has been designed with the prerequisites and requirements of a self-driving vehicle in mind. To achieve the efficient dissemination of updates to all cars the protocol relies on infrastructure-based (cellular) and ad hoc communication (WLAN) between the vehicles. The performance of the protocol is evaluated based on realistic traffic simulations and actual map content.
As our second contribution, we develop and implement an algorithm that detects changes in the road infrastructure (e.g. induced by construction sides) based solely on low-cost sensor information. This detection algorithm facilitates the succeeding update generation of the map data in the identified area. We evaluate the capabilities of the detection algorithm under a real-world data set in the example of a highway construction site scenario.
To enhance the provision of map updates and vehicular sensor data via wireless communication, we conduct our third and most comprehensive contribution. We focus on the design and enhancement of a variety of different techniques and concepts to obtain broad knowledge about the serving wireless network to be provided in a subsequent step as valuable information to related transmission scheduling algorithms. These techniques and concepts include the measurement and prediction of the various performance indicators of actual deployed cellular networks, via low-cost hardware and software, as well as their further usage in simulation and network connectivity maps, always with an emphasis on easy deployability and the reutilization of existing components.
Overall, this thesis presents essential contributions, which in their collectivity support the realization of a robust, dynamic and reliable maintenance cycle of an HD Map for self-driving vehicles.
Typ des Eintrags: | Dissertation | ||||
---|---|---|---|---|---|
Erschienen: | 2020 | ||||
Autor(en): | Jomrich, Florian | ||||
Art des Eintrags: | Erstveröffentlichung | ||||
Titel: | Dynamic Maps for Highly Automated Driving - Generation, Distribution and Provision | ||||
Sprache: | Englisch | ||||
Referenten: | Steinmetz, Prof. Dr. Ralf ; Stavrakakis, Prof. Dr. Ioannis | ||||
Publikationsjahr: | 15 Januar 2020 | ||||
Ort: | Darmstadt | ||||
Datum der mündlichen Prüfung: | 2 Dezember 2019 | ||||
DOI: | 10.25534/tuprints-00009702 | ||||
URL / URN: | https://tuprints.ulb.tu-darmstadt.de/9702 | ||||
Kurzbeschreibung (Abstract): | With an ever-increasing number of vehicles roaming the streets and a general intensification of ongoing daily traffic the current vehicular safety systems are not able to reduce the number of traffic accidents further. As the majority of severe or deadly traffic accidents nowadays is caused by human error, car manufacturers and researchers alike focus on the self-driving vehicle as a promising solution to this problem, as a machine is unaffected from human conditions such as tiredness or drunk driving. To enhance the overall achievable driving safety and comfort the self-driving vehicles rely on an additional map database, besides the hardware sensor system installed onboard. The so-called High Definition Map (HD Map), a highly precise virtual model of the actual real-world provides detailed information about the ongoing traffic situation ahead of the car's sensor ranges. Otherwise critical traffic situations can be resolved by this a priori knowledge and if necessary, a handover of the driving control back to a human driver can be triggered. The maintenance of the HD Map is a major challenge, as due to the importance of the map for the self-driving vehicle map updates have to be realized in much shorter time (minutes instead of months) compared to established concepts common for human-oriented digital navigation maps. This thesis provides contributions in the areas of Distribution, Generation and Provision of such map updates, as the key communication challenges of the maintenance procedure. Our first contribution is the development, implementation and evaluation of a protocol that realizes the context-specific distribution of partial and incremental map updates. The protocol has been designed with the prerequisites and requirements of a self-driving vehicle in mind. To achieve the efficient dissemination of updates to all cars the protocol relies on infrastructure-based (cellular) and ad hoc communication (WLAN) between the vehicles. The performance of the protocol is evaluated based on realistic traffic simulations and actual map content. As our second contribution, we develop and implement an algorithm that detects changes in the road infrastructure (e.g. induced by construction sides) based solely on low-cost sensor information. This detection algorithm facilitates the succeeding update generation of the map data in the identified area. We evaluate the capabilities of the detection algorithm under a real-world data set in the example of a highway construction site scenario. To enhance the provision of map updates and vehicular sensor data via wireless communication, we conduct our third and most comprehensive contribution. We focus on the design and enhancement of a variety of different techniques and concepts to obtain broad knowledge about the serving wireless network to be provided in a subsequent step as valuable information to related transmission scheduling algorithms. These techniques and concepts include the measurement and prediction of the various performance indicators of actual deployed cellular networks, via low-cost hardware and software, as well as their further usage in simulation and network connectivity maps, always with an emphasis on easy deployability and the reutilization of existing components. Overall, this thesis presents essential contributions, which in their collectivity support the realization of a robust, dynamic and reliable maintenance cycle of an HD Map for self-driving vehicles. |
||||
Alternatives oder übersetztes Abstract: |
|
||||
URN: | urn:nbn:de:tuda-tuprints-97022 | ||||
Sachgruppe der Dewey Dezimalklassifikatin (DDC): | 000 Allgemeines, Informatik, Informationswissenschaft > 004 Informatik 600 Technik, Medizin, angewandte Wissenschaften > 600 Technik 600 Technik, Medizin, angewandte Wissenschaften > 620 Ingenieurwissenschaften und Maschinenbau |
||||
Fachbereich(e)/-gebiet(e): | 18 Fachbereich Elektrotechnik und Informationstechnik 18 Fachbereich Elektrotechnik und Informationstechnik > Institut für Datentechnik 18 Fachbereich Elektrotechnik und Informationstechnik > Institut für Datentechnik > Multimedia Kommunikation |
||||
Hinterlegungsdatum: | 26 Jan 2020 20:58 | ||||
Letzte Änderung: | 26 Jan 2020 20:58 | ||||
PPN: | |||||
Referenten: | Steinmetz, Prof. Dr. Ralf ; Stavrakakis, Prof. Dr. Ioannis | ||||
Datum der mündlichen Prüfung / Verteidigung / mdl. Prüfung: | 2 Dezember 2019 | ||||
Export: | |||||
Suche nach Titel in: | TUfind oder in Google |
Frage zum Eintrag |
Optionen (nur für Redakteure)
Redaktionelle Details anzeigen |