PhD opportunities

Real-time abnormal object of behavior detection from LiDAR data

Thesis proposal

Area of expertiseReal-time computer science, robotics,systems and control - Paris
Doctoral SchoolISMME - Systems Engineering, Materials, Mechanics, Energy
SupervisorM. Jean-Emmanuel DESCHAUD
Co-supervisorM. François GOULETTE
Research unitMathematics and Systems
Starting dateOctober 1st 2021
KeywordsLiDAR, deep learning, real time, object detection, autonomous vehicle, train
AbstractThe 'Point Cloud and 3D Modeling' (NPM3D) team of the Center for Robotics at MINES ParisTech / ARMINES is interested in acquisition and processing of 3D point clouds for several applications (3D mapping, mobile robotics, archeology…).
The European 5G-Med project concerns the deployment of various applications using 5G communications, especially for train safety and vehicles. In this context, the NPM3D team, which has just joined the project, has responsible for the design, development and testing of detection solutions real-time objects or abnormal behavior from lidar data, both for automotive and railway applications.
Real-time detection of objects or abnormal behavior is of great importance importance for security issues, and is a difficult problem at the present time unresolved. The use of Lidar data makes it possible to gain in robustness compared to 2D imaging. Recent and promising work has addressed this question using scene flow estimation on 3D data [Mittal et al., CVPR 2020].
ProfileMaster 2 level
Good interpersonal skills, rigor and autonomy
Oral writing and presentation skills
English spoken and written.
(For foreigners) French spoken and written.

Scientific and technological knowledge used:
- 3D data processing
- Image processing
- Machine Learning
Computer developments will be done on PC under Windows or Linux in Python and/or C ++.
FundingFinancement sur programme européen ou multilatéral
PDF document