Skip to Main content Skip to Navigation
Conference papers

Improved dynamic object detection within evidential grids framework

Abstract : The deployment of autonomous robots/vehicles is increasing in several domains. To perform tasks properly, a robot must have a good perception about its environment while detecting dynamic obstacles. Recently, evidential grids have attracted more interest for environment perception since they permit more effective uncertainty handling. The latest studies on evidential grids relied on the use of thresholds for information management e.g. the use of a threshold, for the conflict characterized by the mass of empty set, in order to detect dynamic objects. Nevertheless, the mass of empty set alone is not consistent in some cases. Also, the thresholds used were chosen either arbitrary or tuned manually without any computational method. In this paper, first the conflict is composed of two parameters instead of mass of empty set alone, and dynamic objects detection is performed using a threshold on the evolution of this conflict pair. Secondly, the paper introduces a general threshold along with a mathematical demonstration to compute it which can be used in different dynamic object detection cases. A real-time experiment is performed using the RB1-BASE robot equipped with a RGB-D camera and a laser scanner.
Document type :
Conference papers
Complete list of metadata

Cited literature [15 references]  Display  Hide  Download
Contributor : Nacim Ramdani Connect in order to contact the contributor
Submitted on : Tuesday, October 29, 2019 - 12:23:03 PM
Last modification on : Wednesday, November 3, 2021 - 6:38:26 AM
Long-term archiving on: : Thursday, January 30, 2020 - 12:36:59 PM



Abderraouf Hadj Henni, Angel Soriano, Rafael Lopez, Nacim Ramdani. Improved dynamic object detection within evidential grids framework. 2019 IEEE 15th International Conference on Automation Science and Engineering (CASE), Aug 2019, Vancouver, Canada. pp.1080-1086, ⟨10.1109/COASE.2019.8843016⟩. ⟨hal-02317994⟩



Record views


Files downloads