C++ development of automatic methods for perception around a robotic system on multi-layer LiDAR data• Detection and tracking of dynamic objects in various environments on LiDAR data based on a model free approach and temporal consistency.• Evaluation of the developped algorithm on public datasets such as MD-LiDAR and Semantic KITTI (Obtaining results at the state of the art level on LiDAR data).• Real time semantic segmentation of 3D LiDAR data at 20Hz (algorithm running on CPU only).• Real time tracking of detected objects at 20Hz.Design of an offline, unsupervized method for modeling complex industrial environment (numerical twin) from a LiDAR acquisition and a catolog of CAD models in the scene.

Design of a strategy for decentralized control of a swarm of drones.• Cooperative control law based on Lyapunov stability • Non linear command• Simultaneous attack on a fixed target

Giving weekly oral exams in mathematics to groups of students preparing for national competitive exams of top French engineer schools

Blue collar internship as a construction worker