Skip to Main content Skip to Navigation
Theses

Planification de trajectoire dans un environnement peu contraint et fortement dynamique

Abstract : This thesis studies the trajectory planning of an automated vehicle surrounded by fast moving obstacles in an unconstrained environment (i.e. with no clear lane markings). Two main approaches are proposed: compute a speed profile on a given path, or find a valid path starting from a hypothesis on the speed of the vehicle. The first approach consists of a dynamic adaptation of the speed of an automated vehicle driving in a semi-constrained environment and in the presence of other vehicles. A set of speed profile references is used. They must be compatible with the dynamics of the vehicle and also comfortable for the passengers. Quantitative validations have been conducted in simulation together with qualitative validations on an automated vehicle, which demonstrate the benefits of this planning strategy. In the second approach, the « Reachable Interaction Sets » (RIS) are introduced as a new framework that allows to plan the trajectory of a vehicle surrounded by dynamic obstacles which move faster. The approach removes the temporal aspect of the problem by using a hypothesis about the speed of the vehicle. The remaining problem can be solved out by a static path finding algorithm. Quantitative validations show the advantage of planning approachs based on this framework compared to other state-of-the-art planning strategies.
Complete list of metadatas

Cited literature [169 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-02968172
Contributor : Abes Star :  Contact
Submitted on : Thursday, October 15, 2020 - 2:40:13 PM
Last modification on : Monday, October 19, 2020 - 11:46:20 AM

File

BEAUCORPS_Pierre_De_2019.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-02968172, version 1

Citation

Pierre de Beaucorps. Planification de trajectoire dans un environnement peu contraint et fortement dynamique. Intelligence artificielle [cs.AI]. Sorbonne Université, 2019. Français. ⟨NNT : 2019SORUS082⟩. ⟨tel-02968172⟩

Share

Metrics

Record views

52

Files downloads

11