An Open-Source High-Level Fusion Algorithm in ROS for Automated Driving Applications

Accurate environment perception can be achieved with multiple sensors with different measuring principles - such as camera, lidar and radar sensors. In automated driving applications, this diversity aims, for example, to reduce the susceptibility of sensors to inclement weather. Fusion combines data...

Full description

Saved in:
Bibliographic Details
Published in2022 10th International Conference in Software Engineering Research and Innovation (CONISOFT) pp. 174 - 181
Main Authors Poledna, Yuri, Reway, Fabio, Drechsler, Maikol Funk, Huber, Werner, Icking, Christian, Ribeiro, Eduardo Parente
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.10.2022
Subjects
automated driving
Cameras
Codes
Computer architecture
Data models
environment perception
Sensor fusion
Sensors
Software algorithms
Online AccessGet full text
DOI10.1109/CONISOFT55708.2022.00031

Cover

More Information
Summary:Accurate environment perception can be achieved with multiple sensors with different measuring principles - such as camera, lidar and radar sensors. In automated driving applications, this diversity aims, for example, to reduce the susceptibility of sensors to inclement weather. Fusion combines data from sensors and creates a generic representation of the vehicle's surroundings - which is known as environment model. Based on this, a safe trajectory can be planned and driving functions can actuate accordingly. In this paper, a high-level fusion algorithm is implemented. Its aim is to combine different object lists generated at the sensor-level into a unique global object list. This software is implemented in Python in a ROS workspace and is published as open-source code on GitHub 1 . It was verified that the algorithm can successfully fuse data acquired from simulation-based test drives with sensor models, and create a global object list based on the detections at the sensor-level.
DOI:10.1109/CONISOFT55708.2022.00031