A Monocular Vision-based Specific Person Detection System for Mobile Robot Applications

摘要

The availability and low price of single camera makes them an attractive sensor to develop more sophisticated applications in the field of robotic system. Person detection and tracking systems are important capabilities for applications as a service robot in hospital environment. This work presents a simple method that able to visually detect and track specific person using a single camera based on hybridization method of image information. This method is applied to estimate the position and orientation of a moving target person in crowded hospital environment that can be a nurse or a doctor. The range between the target person and the mobile robot can be computed in real-time using a set of markers so that the robot can control its speed and direction to follow the target person as closely as possible. The experimental results show that proposed algorithm can detect target person under various conditions such as marker features, lighting and in complex backgrounds.