Crossing Road Intersections by an Autonomous Vehicle : Evaluation via Simulation
Résumé
Intersection crossing by an autonomous vehicle (AV) is a difficult task due to multiple situations that may occur in an area with convergence of multiple traffic entities. The vehicle needs to understand its situation with regards to its spatio-temporal relationship with respect to the perceived traffic agents, the context imposed by the intersection geometry, and traffic rules. The testing and evaluation of this manoeuvre in real conditions is difficult even in proving grounds. For this purpose, simulation techniques are used. This paper presents the evaluation of this function using two open-source simulation-AV software stacks, namely Carla-AutoWare & SVL Apollo-Baidu. The purpose is to test their performance using different metrics and as a first step, in this paper the focus is in collision results.
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|---|
| Licence |