Logo Poly Research Laboratory for Computer Graphics & Virtual/Augmented Reality (LIRV) Français ]

Activity Summary

(Back to all activities)

An IoT Platform for Disaster Response

( #réalité augmentée, #réalité virtuelle, #interfaces )

With the use of artificial intelligence and autonomous system, it is now possible to remotely control drones to carry out reconnaissance operations and supervise critical missions (firefighters, rescue, etc.) without endangering humans. The management of this system using a 2D screen limits the display possibilities and can cause confusion in human understanding during the course of a mission. The work consists of designating an innovative solution with current augmented reality technologies.

Interface well adapted to augmented reality

An interface that adapts according to the criticality of the situation, easy to learn and use, since the stress of a person is omnipresent in the course of an emergency mission. The autonomous system includes a set of drones which work in a coherent way (managed by artificial intelligence). The goal of this research work is to design an adapted interface in an emerging technology and to determine its effectiveness on human decision-making. A set of first responder subjects will test the interface prototype while being monitored for attention and stress. We will set up the experiments and collect data, which will be subsequently analyzed to estimate the overall performance of the interface.

Specific objectives :

  1. Understand the functioning of the autonomous system and determine the tools to be produced;
  2. Define and follow a rigorous methodology for the design of the application to develop an adapted interface taking into account the opinions and needs of users;
  3. Validate and perform tests to show the advantage of the interface in real situations during a mission.

"An AR interface to assist human agents during critical missions"

3D Reconstruction and object detection in outdoor environment

The ubiquitous use of Unmanned Aerial Vehicle (UAV), also known as drones, triggered serious risks to population or safety-sensitive infrastructures like airports. To address this concern, robust drone detection systems are required. The goal of our work is to provide the position when a drone appearing on a forbidden zone and the distance from the follower drone.

Due to the various advantages of drones (small size, low price, etc.), the number of the drones expands constantly, making them more common and more accessible. Drones are being used for wide applications like video surveillance, robotics, military, and commercial purposes. However, this pervasive use of drones bring serious risks that need to be addressed.

In this work, we envision an autonomous drone detection system to maintain surveillance and public security. This is one of the first to use computer vision in order to detect and to catch an unauthorized drone. We aim at developing a fully detection system of unauthorized drones using a deep learning based approach. The quantitative experimental results demonstrate that the performance of the detector trained on real or virtual worlds are quit similar.

1-menu.png 2-boussole.png 3-ms-hololens2.jpg 4-magic-leap.jpg 6-batiment_en_feu.png 6-personne_blesses.png 6-simulation_real.jpg detection1.png detection2.png detection3.png detection4.png detection5.png detection6.png detection7.png

Publications :

Highly Qualified Personnel Training (HQP) :

  • Dorra Riahi, September 2018 to August 2019, stagiaire post-doctorale (C)
  • Dany Naser Addin, September 2018 to December 2020, étudiant M.Sc.A. (S)


© 2000-2023 - Polytechnique Montréal - LIRV - Tous droits réservés