Following Control Design of Moving Targets for An Intelligent Vehicle with Skeleton Recognition Tracking

• Main Authors: Hung-Ching Chen, 陳弘清
• Other Authors: Jung-Shan Lin
• Format: Others
• Language: en_US
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/05634336701169538554

碩士 === 國立暨南國際大學 === 電機工程學系 === 104 === This thesis presents the application of an intelligent vehicle for person following purpose in human life. The proposed intelligent vehicle has the potentials to solve the different fields of labor shortage for helping people. This system contains two major research parts, image processing and target following design. In the part of image processing, the system employs Microsoft Kinect to detect the target and uses Microsoft Kinect SDK to obtain the color, depth, and skeleton information of a moving target. The intelligent system applies the gesture of people to determine the target, and utilizes skeleton information of the target for following control design. According to the position of two hip joints, the system obtains the moving target position and stably follows the trajectory of the target. When the target is lost due to some disturbance, the skeleton recognition scheme can apply both color and depth images to restore the direction of target for following in crowded environments. In the part of target following design, this thesis uses PID control to improve the system stability of motor. Fuzzy control is a great way to manipulate the vehicle, because it is similar to logic thinking of person. The system obtains the distance between the target and the vehicle, and then the direction of target for inputs and the speed of vehicle for output. The system uses two fuzzy rule bases to control the behavior of intelligent vehicle, so the system can successfully follow the moving target. As a result, the system can achieve the purpose of person-following with improvement of reliability and robustness under disturbance and complex indoor environments.