Deep Learning in Perception of Autonomous Vehicles

DOI

10.2991/assehr.k.220110.107How to use a DOI?

Keywords

deep learning; autonomous vehicles; self-driving cars; perception; localization; object detection

Abstract

With the development in deep learning and sensor technologies in recent years, the ultimate goal to build full autonomous vehicles (AV) has come closer to practicality. Autonomous vehicles need to be able to percept the environment in order to make the correct decision in controlling the vehicles under different situations. In addition, the process needs to be as accurate as possible since operation under safe conditions is one of the most important issues. Moreover, the efficiency of methods is another crucial factor since complicated traffic requires vehicles to be flexible and react accordingly. Besides, if an AV cannot operate properly and timely, it would be pointless to consider it as an alternative way to the human-control car. While the progress in sensors contributes to the development of AV, the data obtained by sensors is still possible to fail due to environmental or weather conditions. This article first gives a review of the concepts in AV, especially focusing on environmental perception, then concludes and compares several deep learning methods on environment perceptions in the field of AV.

Copyright

© 2022 The Authors. Published by Atlantis Press SARL.

Open Access

This is an open access article under the CC BY-NC license.

Download article (PDF)

TY  - CONF
AU  - Yunxiang Jiang
AU  - Tengyu Hsiao
PY  - 2022
DA  - 2022/01/28
TI  - Deep Learning in Perception of Autonomous Vehicles
BT  - Proceedings of the 2021 International Conference on Public Art and Human Development ( ICPAHD 2021)
PB  - Atlantis Press
SP  - 561
EP  - 565
SN  - 2352-5398
UR  - https://doi.org/10.2991/assehr.k.220110.107
DO  - 10.2991/assehr.k.220110.107
ID  - Jiang2022
ER  -