Path detection for autonomous vehicles using computer vision and on-line learning

• Grant number: 12/21079-1
• Support Opportunities: Scholarships abroad - Research Internship - Master's degree
• Start date: March 01, 2013
• End date: August 31, 2013
• Field of knowledge: Engineering - Mechanical Engineering
• Principal Investigator: Marcelo Becker
• Grantee: Henry Antonio Roncancio Velandia
• Supervisor: Alberto Broggi
• Host Institution: Escola de Engenharia de São Carlos (EESC). Universidade de São Paulo (USP). São Carlos , SP, Brazil
• Institution abroad: Università degli Studi di Parma, Italy
• Associated to the scholarship: 11/03986-9 - Object detection and classification in outdoor environments for autonomous passenger vehicle navigation based on data-fusion of an artificial vision system and laser sensor, BP.MS

Abstract

This research project will be developed on the VisLab at Parma University (Italy) as part of a partnership between LabRoM and VisLab. The main goal of this research is the development of algorithms for path detection using computer vision, especially, for paths in urban and off-road environments. This research aims to improve the current navigation system based on 3D terrain analysis and a priori knowledge of the environment visual characteristics, by adding an online learning procedure. So, we will implement a road feature extractor, a dynamic training module together with a supervised classifier, and a post-processing module for road segmentation and "traversable cost" selection. All the algorithms will be written in C++ and implemented over the GOLD framework, a VisLab tool for intelligent computer vision system design. Several machine learning techniques will be implemented for tuning the classifier. One of the challenges of this project is the design of algorithms aiming a short processing time, since they will be tested online. (AU)

Scientific publications

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

RONCANCIO, HENRY; BECKER, MARCELO; BROGGI, ALBERTO; CATTANI, STEFANO; IEEE. Traversability Analysis Using Terrain Mapping and Online-trained Terrain Type Classifier. 2014 IEEE INTELLIGENT VEHICLES SYMPOSIUM PROCEEDINGS, v. N/A, p. 6-pg., 2014-01-01. (12/21079-1)