Performance Modeling and Parallelization of Environment Perception and Path Planning Algorithms for Heterogeneous Advanced Driver Assistance System Architectures

Beschreibung

Highly automated advanced driver assistance systems (ADAS) will be available for all driving related tasks in future cars. To let such systems become a reality, numerous computationally intensive tasks have to be solved. Traditional single-core processors in automotive electronic control units do not provide enough performance for those tasks. Here, emerging embedded many-core architectures are appealing, such as embedded graphics processing units. Therefore, in this thesis, we evaluate if such architectures can accelerate ADAS algorithms and how algorithms in this context need to be designed to take full advantage of these architectures. Hereby, special attention is paid to the parallelization of environment maps and path planning algorithms as well as the corresponding performance models.

Autor*in

Jörg Fickenscher

Themen in »Performance Modeling and Parallelization of Environment Perception and Path Planning Algorithms for Heterogeneous Advanced Driver Assistance System Architectures«

Advanced Driver Assistance Systems GPU Parallelization Environment Perception and Path Planning

Stimmen zu »Performance Modeling and Parallelization of Environment Perception and Path Planning Algorithms for Heterogeneous Advanced Driver Assistance System Architectures«

Details

Link teilen