Pedestrian detection for inteligent vehicles

• Grant number: 12/19867-1
• Support Opportunities: Scholarships in Brazil - Doctorate
• Start date: January 01, 2013
• End date: December 31, 2015
• Field of knowledge: Physical Sciences and Mathematics - Computer Science
• Principal Investigator: Denis Fernando Wolf
• Grantee: Francisco Alexandre Ribeiro de Alencar
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil

Abstract

One of the primordial problems is the robotic autonomous navigation that even after decades of study, still presents great opportunities for scientific research. Currently, a major focus of research in autonomous navigation is the obstacles identification in outdoor and unstructured (urban or agricultural), due to the great possibilities of application. In such an environment unstructured need is the development of intelligent autonomous vehicles to avoid obstacles that may appear in the path, or even take paths not provided.In this context the task of identifying not only the existence of an obstacle, but how this obstacle can behave (direction, speed and other characteristics) is of utmost importance for an autonomous vehicle, because from these data the system can take important decisions how to behave during your commute (which trajectory and speed are necessary).The purpose of this research plan is to develop a system of identification of dynamic obstacles (people, vehicles or other moving obstacles) from the capture stereo images to aid autonomous navigation. The proposed system focuses primarily on the identification of people and should be able to do it in real time, in order to enable the navigation system to meet not only the distance and position, as well as the possible trajectory of the obstacle from the vehicle , thus taking the necessary decisions (steering, braking and acceleration).