Fraunhofer-Gesellschaft

Publica

Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Towards Collaborative Perception in Automated Driving: Combining Vehicle and Infrastructure Perspectives

Paper presented at CARS 2021, 6th International Workshop on Critical Automotive Applications: Robustness & Safety, September 13, 2021, Virtual
 
: Zacchi, Joao-Vitor; Trapp, Mario

:
Fulltext urn:nbn:de:0011-n-6422206 (264 KByte PDF)
MD5 Fingerprint: 05a04188256de029b337f866b28f7484
Created on: 21.10.2021


2021, Paper hal-03366395 (hal.archives-ouvertes.fr), 4 pp.
International Workshop on Critical Automotive Applications - Robustness & Safety (CARS) <6, 2021, Online>
European Commission EC
H2020; 812788; SAS
Safer Autonomous Systems
English
Presentation, Electronic Publication
Fraunhofer IKS ()
collaborative perception; automated driving system; Bayesian occupancy filter

Abstract
Environment perception constitutes a foundational block for autonomous systems such as automated driving systems. Enhancing such features is imperative to breach the barrier of complex environments such as urban scenarios. Occlusions, appearances, and disappearances are a few of the difficulties traditional tracking algorithms may face in an urban context that hinders their performance. Moreover, approaches that deal with the data association problem are still physically limited by the point-of-view of the ego vehicle. In order to address these issues, we propose in this position paper a framework to merge different perspectives enabling collaborative perception and thus to enhance the dependability of the environment perception of automated vehicles in complex scenarios. To this end, each participant, i.e., automated vehicles and infrastructure, sends their perception results to the framework. A perception result includes Bayesian Occupancy Filter providing probabilistic information about object positions. Moreover, the results might include an additional classification of the objects, enabling us to optimize predicting future trajectories of the objects, which is particularly important for non-automated participants such as human-driven cars or pedestrians. The framework facilitates a more complete and clarified view of the context to enhance decision-making of the individual vehicles.

: http://publica.fraunhofer.de/documents/N-642220.html