IngÃnieur de Recherche - computer vision / computer graphics

Position and assignments

The URBAN VISION project is a 3 year research project that concerns a consortium of research and higher learning establishments on the campus of the Paris-Est University, including Ecole des Ponts ParisTech and Université Gustave Eiffel. The objective of this project is to bring together Computer Vision, Virtual Reality and Human Vision in the context of transports in modern cities. Applications include driving simulation, autonomous vehicle solutions development and testing, human vision research with respect to traffic, visibility and weather.

In the context of the URBAN VISION Project, the candidate will be appointed for a duration of 12 months. This post-doc focuses on the acquisition, post-processing and surface reconstruction of point clouds produced by Lidar acquisitions. A first acquisition has been done with a calibrated mobile mapping vehicle equipped with a geopositioning system, Lidar sensors and RGB cameras. This massive and accurate dataset will serve as a reference dataset to reconstruct a full and accurate 3D surface model, to be compared with the surface reconstructed from a point cloud acquired by an innovative alternative acquisition system based on a solid state Lidar (developed by Ouster). The post-doc will focus on producing and comparing accurate surface reconstructions produced by state-of-the-art algorithms on both the reference and alternative point cloud datasets. The alternative solution, which is significantly less expensive, is more likely to be used in a commercial context and offers an interesting novel measurement of the ambient NIR lighting. However, given the relative novelty of this acquisition setup, a reasonable adjustment period will be needed to validate the acquisition pipeline. The post-doc will have to build a new acquisition set-up and strategy in order to produce highly accurate 3D models, to evaluate the potential of this novel Lidar sensor.

Responsibilities

• Processing and analysis of 3D point cloud
• Supervising 3D acquisition with lidar solid state Ouster
• Deploying a full 3D workflow reconstruction on the acquisitions
• Testing these solutions in the context of building large scale realistic models of city blocks
• Participating in the experimental development and testing of one or several applications of these models
• Making the 3D models manageable for virtual reality applications
• Contributing to scientific publications in international conferences and journals

Duration and salary

Salary : 2000 ? approximately

Geographic mobility

No business trip

Start date

Profile

• PhD in physics, mathematics, computer science or equivalent
• Working with 3D points cloud
• Excellent knowledge in C++
• Experience with Apache / Spark is appreciated
• Experience with ROS is appreciated

Goals

In the context of the URBAN VISION Project, the candidate will be appointed for a duration of 16 months. His/her main activity is the integration of a software pipeline which goes from the capture of images in a city to the virtual experience of walking/driving in this same city. It includes the following steps:

• Processing and analysis of 3D point cloud
• Supervising 3D acquisition with lidar solid state Ouster
• Deploying a full 3D workflow reconstruction on the acquisitions

Location

The candidate will work on IGN campus at Saint-Mande and on the campus of Marne-la-Vallee (20 mn from Paris, depicted below).