PORTALE DELLA DIDATTICA

Ricerca CERCA
  KEYWORD

GR-09 - GRAphics and INtelligent Systems - GRAINS

Pedestrian behaviour characterization for Automated Driving (AD) and Driver Assistance (DA) systems

azienda Thesis in external company    


Reference persons FABRIZIO LAMBERTI

Research Groups GR-09 - GRAphics and INtelligent Systems - GRAINS

Thesis type TESI IN AZIENDA

Description A key functionality to support Automated Driving (AD) and Driver Assistance (DA) systems is the ability to detect obstacles in order to constantly adapt the vehicle trajectory to a changing environment. The overall objective of this project is to develop an Artificial Intelligence module to detect and characterize obstacles, and in particular pedestrians, from sensors mounted on board of vehicles: this module should be able to detect and track pedestrians, classify their actions (walking, crossing the street) and predict possible trajectories, along with their level of confidence. The activities to be carried out within this thesis include the generation of suitable training data, leveraging public datasets but also simulation environments such as Carla or AirSim. The expected outcome is a proof of concept of the main capabilities of the module, by implementing and training deep learning models for the main functionalities (detection, action classification, trajectory modeling); the main input stream considered will be from onboard RGB cameras. The activities will be carried out at GRAINS lab, in collaboration with Centro Ricerche Fiat (FCA). The thesis will leverage existing resources at GRAINS (pre-trained models for action detection, classification, and tracking) and VR@Polito labs (computer graphics). Strong programming and analytical skills are required. Prior knowledge of machine learning / deep learning / computer vision and practical experience with Keras/Tensorflow are preferable.

See also  http://grains.polito.it/work.php


Deadline 06/03/2020      PROPONI LA TUA CANDIDATURA




© Politecnico di Torino
Corso Duca degli Abruzzi, 24 - 10129 Torino, ITALY
Contatti