A Runtime-Efficient Multi-Object Tracking Approach for Automotive Perception Systems

There is an increasing demand for fully autonomous driving with the spread of advanced driver assistance systems. However, a higher level of automation requires an enhanced environment perception system. The automotive smart sensors detecting the surrounding objects are usually subject to various un...

Full description

Saved in:
Bibliographic Details
Published in:SACI (International Symposium on Applied Computational Intelligence and Informatics. Online) pp. 000785 - 000792
Main Authors: Lindenmaier, Laszlo, Czibere, Balazs, Aradi, Szilard, Becsi, Tamas
Format: Conference Proceeding
Language:English
Published: IEEE 23.05.2023
Subjects:
Automation
Automotive applications
Computational complexity
Environment perception
Informatics
Multi-object tracking
Nearest neighbor methods
Noise measurement
Object detection
Real-time systems
Uncertainty
ISSN:2765-818X
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:There is an increasing demand for fully autonomous driving with the spread of advanced driver assistance systems. However, a higher level of automation requires an enhanced environment perception system. The automotive smart sensors detecting the surrounding objects are usually subject to various uncertainties. Objects are sometimes misdetected, false detections may also occur, and the sensor measurements are noisy. Multi-object tracking algorithms aim to compensate for these uncertainties, estimating precisely and continuously the state of surrounding objects. The real-time execution of these algorithms is crucial in automotive applications. This paper presents an improved runtime-efficient local nearest neighbor-based approach. The performance metrics of the proposed method get close to the state-of-the-art algorithms, besides having significantly lower computational complexity.
ISSN:2765-818X
DOI:10.1109/SACI58269.2023.10158542