La société Gambi-M et le laboratoire LIRMM UMR CNRS 5506 recrutent un doctorant dans le cadre d'une bourse CIFRE pour travailler sur l'inspection automatique d'un bâtiment industriel à l'aide d'un drone.
Le travail s'effectuera à plein-temps dans les locaux du laboratoire avec des déplacements très ponctuels des acteurs sur le site de l'entreprise.
Date limite : 1 aout 2018
This research proposal (CIFRE thesis) is targeted for MSC students specialized in robotics, electronics or control engineering. The selected candidate will work in the highly promising field of autonomous flying vehicles on cutting edge technologies and research innovations, within full collaboration between private (Gambi-M) and public sectors (LIRMM UMR CNRS 5506).
Requirements and qualifications:
The candidate would have the following skills:
Starting date:September, 2018
Company: Gambi-M, Impasse de l’Hermitage, 30200 Bagnols sur Cèze - FRANCE
Laboratory: Laboratoire d’Informatique et de Microélectronique de Montpellier – Université de Montpellier, école doctorale I2S.
The work will mainly take place in Montpellier with some meetings in Bagnols sur Cèze.
To have a dense and accurate 3D model of an industrial building at disposal is really important for safety procedures, inspections, maintenance tasks or dismantling…. Such a 3D virtual model gathers all relevant technical data and eases the monitoring of the building during its lifecycle until its eventual dismantling. The making of this model is often done manually by operators who have to do repetitive, time consuming and tedious measurement tasks due to the huge size of the buildings.
Nowadays, we think that drones have tremendous potential to achieve this task since they are becoming relatively inexpensive, with extended battery life. Moreover, they are able to embed many sensors. We would like to have this task done by an aerial robot which navigates as autonomously as possible in buildings whose structure is unknown.
Within this prospect, we propose to decompose the project in two main steps. First, a navigation using SLAM techniques will be done for obtaining a rough 3D model of the inspected area. Then, for reducing the acquisition time, we will compute the shortest path in this sparse 3D model for a second inspection which ensure the best 3D reconstruction by photogrammetry techniques.
This research project follows preliminary works done during a 5 months Msc internship in robotics which started in March 2018. This work enabled us to evaluate the feasibility of using a commercial drone (Mavic Pro, from DJI, ARDrone from Parrot), as well as to initiate future experiments.
One of the company's interest targets the optimization of digital surveying for indoor facilities. We want to develop an aerial robot able to navigate in an unknown environment without GPS using vision and laser sensors.
For this project, experimentations will be led in a full-scale copy of an industrial room, in Gambi-M’s facility.
This project aims at:
The salary will highly depend on experience and effective skills of the successful candidate.
Hornung, A., Wurm, K. M., Bennewitz, M., Stachniss, C., & Burgard, W. (2013). OctoMap: An efficient probabilistic 3D mapping framework based on octrees. Autonomous Robots, 34(3), 189-206
MUR-ARTAL, Raul et TARDÓS, Juan D. Orb-slam2: An open-source slam system for monocular, stereo, and rgb-d cameras. IEEE Transactions on Robotics, 2017, vol. 33, no 5, p. 1255-1262.
L. von Stumberg, V. Usenko, J. Engel, J. Stückler and D. Cremers, From monocular SLAM to autonomous drone exploration, 2017 European Conference on Mobile Robots (ECMR), Paris, 2017,
Grzonka, S., Grisetti, G., & Burgard, W. (2012). A fully autonomous indoor quadrotor. IEEE Transactions on Robotics, 28(1), 90-100.
ICRA 2018 Workshop on Aerial Robotic Inspection and Maintenance: Research Challenges, Field Experience and Industry Needs
Thomas Chaperon. Segmentation de nuage de points 3D pour la modélisation automatique d'environnements industriels numérisés. Modélisation et simulation. École Nationale Supérieure des Mines de Paris, 2002. Français.
(c) GdR 720 ISIS - CNRS - 2011-2018.