PARallele Implementierungs-Strategien für das Hochautomatisierte Fahren
The goal of the project PARIS is the development of a new highly parallel processing unit with optimized cores for the calculation of complex and computationally intensive fusion algorithms. A major field of these algorithms is machine learning, e.g. for computer vision, but also for behaviour generation in the context of automated driving.
For the development of these algorithms, it is essential to create programming guidelines and verification methods already during the development process, e.g. by using virtual prototyping methods. At the end of the project, all algorithms will be integrated in a test vehicle and demonstrated in a real driving scene.
The Institute for Automotive Engineering - RWTH Aachen University takes the role of the vehicle-related function developer in the project ELEVATE PARIS and is also responsible for the verification and validation of the designed processing unit at the end of the project.
Beside the definition of the use case(s) and thereof derived requirements for the hardware, all criteria for evaluation, verification and validation are defined at the beginning of the project.
A further work package for ika is the evaluation of existing and newly developed machine learning algorithms in the automobile context, but especially the development of a highly automated driving function, which will be applied to ika's test vehicle (the function consists of environment perception and interpretation, behaviour generation, trajectory planning and vehicle control).
The new processing unit will be evaluated regarding its processing power and efficiency for executing complex algorithms in conclusion of the project.
04/2017 – 03/1000
NISYS GmbH, Silexica GmbH, Ruhr-Universität Bochum, BASELABS GmbH, Leibniz Universität Hannover, videantis GmbH, Robert Bosch GmbH, RWTH Aachen, Elektrobit Automotive GmbH