Cooperative people detection and tracking strategies with a mobile robot and wall mounted cameras

Alhayat Ali Mekonnen 1
1 LAAS-RAP - Équipe Robotique, Action et Perception
LAAS - Laboratoire d'analyse et d'architecture des systèmes [Toulouse]
Abstract : This thesis deals with detection and tracking of people in a surveilled public place. It proposes to include a mobile robot in classical surveillance systems that are based on environment fixed sensors. The mobile robot brings about two important benefits: (1) it acts as a mobile sensor with perception capabilities, and (2) it can be used as means of action for service provision. In this context, as a first contribution, it presents an optimized visual people detector based on Binary Integer Programming that explicitly takes the computational demand stipulated into consideration. A set of homogeneous and heterogeneous pool of features are investigated under this framework, thoroughly tested and compared with the state-of-the-art detectors. The experimental results clearly highlight the improvements the different detectors learned with this framework bring to the table including its effect on the robot's reactivity during on-line missions. As a second contribution, the thesis proposes and validates a cooperative framework to fuse information from wall mounted cameras and sensors on the mobile robot to better track people in the vicinity. Finally, we demonstrate the improvements brought by the developed perceptual modalities by deploying them on our robotic platform and illustrating the robot's ability to perceive people in supposed public areas and respect their personal space during navigation.
Complete list of metadatas

Cited literature [196 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-01068355
Contributor : Emilie Marchand <>
Submitted on : Thursday, September 25, 2014 - 2:53:15 PM
Last modification on : Friday, June 14, 2019 - 6:31:05 PM
Long-term archiving on : Friday, December 26, 2014 - 10:55:26 AM

Identifiers

  • HAL Id : tel-01068355, version 1

Citation

Alhayat Ali Mekonnen. Cooperative people detection and tracking strategies with a mobile robot and wall mounted cameras. Robotics [cs.RO]. Université Paul Sabatier - Toulouse III, 2014. English. ⟨tel-01068355⟩

Share

Metrics

Record views

525

Files downloads

1208