A Human-Following NAO Robot Using Python Programming Language

Robots are today studied by a growing number of research institutes to research of putting robots, especially humanoid robots, into the market that to replace traditionally human workers. Apart from this, robot is a powerful tool to engage students with developing AI in education. In the future, robots will be much more human-friendly with human beings and it is a trend to enable robots to co-exist with human beings. With more and more humanoid robots being manufactured, robots and humans are getting as close as possible to each other. Humanoid robots are appearing in many areas of life, such as doing housework, taking care of the elders, and accompanying children. To create a healthy co-existence environment of human and robot, robots have to learn to follow mankind as one of the methods to move. This thesis presented face and object detection to achieve human-following function. The NAO robot could avoid obstacles and move to human beings. This project included Google speech recognition, which converted audio files to text files and enabled the NAO robot to carry out a simplified communication with human beings, using OpenCV trained cascade, which recognized human’s face and the corresponding box. Also, the Caffe model, which drew face frames and face’s confidence level was used. Finally, the work utilized NAOqi modules to control the whole-body joint of the NAO robot and employ the NAO robot to follow a human. The study was carried out with a pre-trained model to detect the human’s face, walking to the front of human beings. In addition, the NAO humanoid robot would avoid a specific box and follow the user. The result indicated that OpenCV detected face by using an object or a face trained cascade file is not accurate enough at the moment. However, it is enough for this thesis