Object detection with radar : present and future automotive technology
Correu electrònic de l'autordanidiaz0024gmail.com
Realitzat a/ambHochschule für Technik und Wirtschaft Dresden
Tipus de documentTreball Final de Grau
Data2022-07-20
Condicions d'accésAccés obert
Llevat que s'hi indiqui el contrari, els
continguts d'aquesta obra estan subjectes a la llicència de Creative Commons
:
Reconeixement-NoComercial 4.0 Internacional
Abstract
Radar-based object detection in cars is an integral part of autonomous driving systems. Radar
sensors benefit from their excellent robustness in adverse weather conditions such as snow, fog
or heavy rain.
Although machine learning-based object detection is traditionally a camera-based domain, great
progress has been made in lidar sensors, and radar is also catching up.
Radar has been a key element of advanced automotive driver assistance systems for more than
two decades.
As an inexpensive, all-weather, long-range sensor that simultaneously provides speed
measurements, radar is expected to be indispensable for the future of autonomous driving.
Traditional radar signal processing techniques are often unable to distinguish reflections from
objects of interest and are generally limited to detecting the peaks of the received signal.
These peak detection methods convert the radar signal as an image into a sparse point cloud.
Fully autonomous vehicles and the need to improve road safety have increased the reliability
requirements of various advanced driver assistance systems (ADAS). Automotive radar is a key
component of ADAS, as it adds safety and comfort features to vehicles.
One of the main challenges in developing automotive radar is to demonstrate its reliability,
especially in the most difficult cases. Building and testing radar systems for specific cases is time-
consuming, costly and impractical. Simulation is the only practical way to investigate the
countless practical cases of automotive radar.
One interesting case is the reduction of radar returns due to sharp road curves. In particular,
crucial targets with low radar cross sections (RCS), such as pedestrians, can become invisible to
radar when driving on sharp curves.
This paper will implement a radar system for the simulation of object detection of a vehicle, and
aims to show and analyse how reliable such systems can be, as well as their problems and more.
TitulacióGRAU EN ENGINYERIA D'AUTOMOCIÓ (Pla 2017)
Fitxers | Descripció | Mida | Format | Visualitza |
---|---|---|---|---|
APPENDIX.pdf | 14,62Mb | Visualitza/Obre | ||
DanielDiaz_FinalThesis.pdf | 7,377Mb | Visualitza/Obre |