Description
Date depot: 14 février 2023
Titre: Motion and Occupancy Forecasting for Autonomous Driving
Directeur de thèse:
Matthieu CORD (ISIR (EDITE))
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Images et vision
Resumé: Autonomous robotics, and autonomous driving (AD) as a special case, are at the forefront of such problems. Arguably close to delivering major breakthroughs to a number of industries and to our society at large, these technologies are facing an immense challenge: How to decipher, in real-time, very complex and rapidly changing environments to take decisions accordingly such that prescribed goals are attained efficiently and safely? Answering this question requires the design of a system that
constantly performs sensing, predicting, planning, and action. At several stages, temporal prediction is needed: to pass key information to the next stage, to exploit at best information from the past, and to make good decisions. The nature of the parameters to predict and the temporal horizon of the prediction vary across the system.
The aim of this thesis is to focus on the short-term prediction of some aspects of the surrounding environment of a driverless car. The environment is received as a “scene” by the different sensors of the vehicle, a visual scene for the cameras, a 3D point cloud for the laser scanners (LiDAR), orientation and speed for the IMU, absolute coordinate for the GPS, etc.
Doctorant.e: Chambon Loick