Eventos Anais de eventos
CONEM 2022
XI Congresso Nacional de Engenharia Mecânica - CONEM 2022
ADVANTAGES OF SENSORS AGGREGATION IN AUTONOMOUS CARS
Submission Author:
Albano Leo Ehrenbrink , RJ
Co-Authors:
Albano Leo Ehrenbrink, Max Suell Dutra
Presenter: Albano Leo Ehrenbrink
doi://10.26678/ABCM.CONEM2022.CON22-0324
Abstract
This article studies the advantages of using several sensors in autonomous vehicles. Nonetheless, the variation of the uncertainties of different sensors in different scenarios, mainly in the transition between scenarios, causes a decrease in the reliability of the decision-making model. This work shows the advantages and disadvantages of each type of sensor typically used in autonomous car projects, the evolution of assistant drive systems, and an understanding of how the liability increases overtime for these systems. This work also makes a proposition how aggregate sensors inputs for a better interpretation of the scenario with the consequent increase in the robustness of the decision-making model. Finally, this article also highlights the importance of sensor aggregation mainly in everyday situations that present significant challenges to computer vision as well as systems that use: LiDAR (laser imaging, detection, and ranging), RGBD (Red, Green, Blue, and Distance), or stereo vision to determine the distance of objects to estimate the vehicle location known as SLAM (Simultaneous localization and mapping). The solution for these challenges includes the benefit of aggregation of sensors that reads different wavelengths and are susceptible to various obstacles limitations. Despite this, the interpretation of the car's behavior or location can contribute to choosing what sensor is more appropriate to increase the reliability of the decision-making model.
Keywords
sensor aggregation, Autonomous vehicles, Computer Vision, LiDAR, SLAM
