3D digital technologies play a key role in our society today, and they are essential for preserving, promoting and understanding the historical and cultural heritage. In particular, 3D laser scanners (Lidars) allow to analyze the surface of complex structures (e.g. the facade of a building) and to extract detailed information about its geometry and appearance. The laser measurements are traditionally processed offline to build a 3D virtual model for the end-user. However, in spite of recent technological advances, the 3D reconstruction pipeline (from point-data collection to 3D modeling) is far from being automatic and it requires a high degree of human intervention, which has a strong impact on the costs and on the length of a measurement campaign. Other challenging issues are the generation of homogeneous and complete models, the presence of occlusions, and the exposure to accident risk for the personnel.
The goal of the ScanBot project is to develop and to experimentally validate new 3D mapping, cooperative control and active perception algorithms for mobile robotic platforms, to make the data collection and processing phases of the 3D reconstruction pipeline, fully automatic. The coordination of robotic platforms, called « ScanBots », will be instrumental in speeding up the acquisition process and improving the accuracy and uniformity of the 3D model, while ensuring the safety of the personnel. Heterogeneous vehicles and sensors (robots equipped with Lidars on the ground, and drones with on-board cameras aloft) will provide complementary information about the environment to be reconstructed. The measurements gathered by each robot will be used to navigate autonomously and safely, and also to generate a 3D local model of the environment. The ensemble of the local maps will be automatically registered and merged into a global coherent model.
Lasergrammetry; Computer Vision; Mobile Robots; Active Perception; Cooperative Control
Profile of the ideal candidate:
We are seeking an outstanding and enthusiastic candidate with a strong background in computer vision, robot perception, control theory and optimization. Excellent programming skills (C, C++, Matlab) and a good knowledge of the most common computer vision libraries and software tools for point cloud visualization and processing, are also requested. Applicants should have a MSc in Computer Engineering, Electrical Engineering, Control Engineering or closely related fields.
The candidates should send their CV, list of grades, research experience, language skills and motivation letter as a single pdf to Dr. Fabio Morbidi, firstname.lastname@example.org (please use [PhD position ONERA/UPJV] as subject line of your email). The selected candidates will be invited for an interview via Skype or telephone before June 15, 2018. Questions on the position can be addressed to Dr. Fabio Morbidi (email@example.com) or to Dr. Julien Moras (firstname.lastname@example.org). If you are planning to attend ICRA’18 in Brisbane, a job interview can be arranged with Dr. Fabio Morbidi.
Expected start date:
October 1st, 2018
The « Office National d'Etudes et de Recherches Aerospatiales » (ONERA) is the French national aerospace research center. It is a public establishment with industrial and commercial operations, which carries out application-oriented research to support innovation and competitiveness in the aerospace and defense sectors. ONERA was created in 1946 as « Office National d’Etudes et de Recherches Aeronautiques ». The DTIS (Department of Information Processing and Systems) is located in Palaiseau, in the southern suburbs of Paris, about 17 km from the center of Paris (www.onera.fr/dtis/traitement-information-et-systemes).
The University of Picardie Jules Verne (UPJV) is a public university located in Amiens, France (120 km north of Paris). It was founded in 1969 and it is organized in 12 faculties, 35 research units and 6 campuses in the Hauts-de-France region. In 2016, 28000 students enrolled at UPJV. With a personnel of 80 people (40 permanents, 36 Ph.D. students and post-docs, and 4 administrative and technical staff), the MIS laboratory is the leading laboratory of UPJV in the area of Information and Communication Technologies (www.mis.u-picardie.fr). The Robotic Perception group is internationally recognized for its research on non-conventional vision for mobile robotics and large-scale 3D modeling for digital heritage (for more information, visit the web page of the E-cathedral program: https://mis.u-picardie.fr/e-cathedrale/).