IMT Atlantique, internationally recognized for the quality of its research, is a leading general engineering school under the aegis of the Ministry of Industry and Digital Technology, ranked in the three main international rankings (THE, SHANGHAI, QS).
Located on three campuses, Brest, Nantes and Rennes, IMT Atlantique aims to combine digital technology and energy to transform society and industry through training, research and innovation. It aims to be the leading French higher education and research institution in this field on an international scale. With 290 researchers and permanent lecturers, 1000 publications and 18 M€ of contracts, it supervises 2300 students each year and its training courses are based on cutting-edge research carried out within 6 joint research units: GEPEA, IRISA, LATIM, LABSTICC, LS2N and SUBATECH.
The postdoctorate will be based in the Brest campus of IMT Atlantique joining team RAMBO of CNRS Lab-STICC, UMR 6285 and will be cosupervised by Ass. Prof Panagiotis Papadakis (team coleader) and Prof. Guillaume Moreau of team INUIT. The research fields of RAMBO team encompass cognitive robotics, robot learning, ambient assisted living and human-robot interaction while the group possesses state-of-the-art research facilities and equipment, notably a LivingLab in eHealth as well as diverse service/field robots and manipulators.
This position is offered in the context of the project LEASARD - Low-Energy deep neural networks for Autonomous Search-And-Rescue Drones – financed by CominLabs and is a collaborative project between research teams 2AI, RAMBO and INUIT of Lab-STICC in Brest, LS2N / Armen in Nantes and IRL Crossing in Australia. The purpose of the project is to develop optimized deep neural networks for vision and control of drones, by capitalizing on recent advances on event-cameras coupled with colour cameras and dedicated FPGA hardware for embedded processing.
The postdoctoral researcher will undertake research in the field of deep-learning for vision and control of an aerial drone destined to operated in urban search and rescue (USAR) missions. Of particular interest will be the development of deep neural networks using event-camera images alongside conventional colour images so as to favour network sparcity and complementarity when implemented on dedicated FPGAs capable of on-demand reprogramming. The goal thus consists in advancing the state-of-the-art in terms of efficiency of deep neural networks used during the drone mission for vision and control, so as to increase the operational autonomy while moderating the impact in the overall efficacy. To this end, the paradigms of either end-to-end (mapless) or conventional SLAM-based navigation can be investigated depending on expected benefits and the scenario constraints. The development of algorithms will be initially sought via simulation tools and environments for drone navigation and progressively transferred to the real sensors and hardware.
The postdoctoral researcher will work in close collaboration with project partners. Upon joint agreement, the 1st year of postdoc can be concatenated with an additional 1-year postdoc at the LS2N/Armen team in Nantes, under the cosupervision of Dr. Isabelle Fantoni. The project provides necessary funding to cover publication, mobility and working environment expenses.
Details on the application procedure and requirements are available via the following link :
Keywords : event-cameras, event-based vision, deep-learning, drones, usar
(c) GdR 720 ISIS - CNRS - 2011-2022.