People detecting and tracking using laser and vision

As part of design of a personal robot for operation in an everyday environment thereis need to endow the system with facilities to track a person as it is taken on a tour of theenvironment. In the literature a number of different methods based on laser tracking and computer vision have been presented. However, most of these methods are not particularly robust and in many cases the methods do not operate in realtime. In this master thesis a system for people detecting and tracking has been implemented using laser and vision. The information given by both scanners are used for two different purposes, laser range data are used to detect persons and the images grabbed by the camera are used to confirm the hypotheses made by the laser. Methods for people tracking based on laser and vision have two main problems. The ones based on laser are not very robust and the ones based on vision hardly ever operate in real time. This project is aimed at taking the main advantages of both methods: Laser Advantage: High Speed => Works in real time Disadvantage: Not robust => Not reliable 100% Vision: Advantage: Robust => More reliable Disadvantage: Low speed => Does not work in real tim