Portable implementationsfor heterogeneous hardwareplatforms in autonomousdriving systems

verfasst von

Oliver Jakob Arndt, Parwesh Rallapalli, Holger Christoph Blume

Abstract

Complex driver-assistance systems that analyze driving situations based on a range of sensors enable autonomous driving vehicles-a key aspect of smart cities. This massive automation necessitates computationally powerful and energy-efficient hardware devices available in each individual driving unit. Heterogeneous multiprocessor system-on-chips provide excellent performance-to-power characteristics for the use in driver-assistance applications. Since these programmable chips use flexible software, they theoretically feature high maintainability and portability. However, due to the lack of programmability of different parallel and heterogeneous processing units, developers can barely fully exploit all computational capabilities. To overcome the gap between theoretical peak performance and the effectively gained speedup, diverse programming approaches and supportive tools have emerged. This work presents an overview of the most important trends and contributes a middleware approach for abstracting, and thus unifying, the programming for homogeneous and heterogeneous architectures.

Organisationseinheit(en)

Institut für Mikroelektronische Systeme

Typ

Beitrag in Buch/Sammelwerk

Seiten

113-143

Anzahl der Seiten

31

Publikationsdatum

19.07.2019

Publikationsstatus

Veröffentlicht

ASJC Scopus Sachgebiete

Sozialwissenschaften (insg.)

Ziele für nachhaltige Entwicklung

SDG 11 – Nachhaltige Städte und Gemeinschaften

Elektronische Version(en)

https://doi.org/10.1016/B978-0-12-816637-6.00006-3 (Zugang: Geschlossen)