Research And Implementation Of Intelligent Recognition Robot Based On WiFi Wireless Network

The development of computer science and other subjects have promoted rapid progress in robotic technology, also home service robot market growth in robot industry. Nowadays, robotic technology is widely used in many houses on earth, not just in laboratory or factory. There's a big market in family robot industry, so that more and more researchers focus on robotic related subjects, such as artificial intelligence, communication technology, pattern recognition.Aiming at the huge family robot market, against the problem of higher costs in traditional robot industry, the goal of this thesis is to design and build a WiFi wireless network based intelligent recognition robotic system.Firstly, we deeply analyze the development status of intelligent recognition robot industry, and conclude the major problem, opportunity, challenge we may face in this area by reading related literatures. Recently, for widespread personal computers and the Internet, family used wireless networks are very cheap to build. The system which is composed of robots, PCs and wireless router fully makes use of the hardware that may exist in house environment. In this system, robot only takes the responsibility of mechanical operation, data acquisition and communication. But the operation such as character recognition and face detection which needs complex computing is executed on PC. Communication between robot and PC is taken through WiFi network.In order to realize the function of character recognition, in this thesis, we have especially design an algorithm for recognition of the characters in an image. The algorithm is divided into two parts, including detection and extraction of text from image and off-line character recognition. We elaborate every step of the algorithm in this thesis. Besides, we analyze and compare several common algorithms for edge detection. At the end, experiments have been carried out, and the recognition accuracy is 97%.For the function of autonomous patrol, we have designed the human face detection and smile recognition algorithm. Haar-like feature is extracted , and AdaBoost Learning is used to train an cascade classifier in face detection. We use the Gabor filter to extract feature and use AdBoost algorithm to train Single-Stump weak classifier to finally build a strong classifier. The human-computer interface is designed using C++ language and WTL(Windows Template Library)framework. With our system, operators can control a mobile robot to navigate in the laboratory and watch the real time video transmitted back from the camera in front of the robot. Also, the robot can be set into autonomous patrol or character recognition mode.