| The quickly development of the modern robot technology has resulted in itself a more and more mature application in today's engineering community. As a high-tech production which integration mobile robotics, computer technology, bionics, pattern recognition etc, the intelligent wheelchair's appear has draws more and more attention. Meanwhile, with its function like maps memory and communication with its user through man-machine interface, the wheelchair has brought a revolution in this filed. Although it's used some technology from the mobile robotics, the wheelchair is used to provide service for people, so the prime function is to keep safe for its user. Planning a collision free path which has no collision with other object automatic or half-automatic is the intelligent wheelchair's prime task. Also the autonomous navigation is a hot topic today.Safety navigation based on visual senior is an advanced technology nowadays. Compared with other navigation methods, it has some advantage like enough information for calculate; wide in detect scope; integrity for object information etc. The study platform in this paper is the intelligent wheelchair, which is also the main application object. For this paper, the main contribution is as following:1. Proposed a new method for camera calibration-the BP neural network camera calibration. As a first step for the vision system, the accuracy of camera calibration have a great effect on the reliability of the system work. After a whole analysis for the traditional methods of camera calibration, and point out the limitation of this methods, then it combine the modern neural network technology and proposed the BP neural network for camera calibration. In order to prove the effective of this method, the paper built a three layer BP neural network mode, and using the mode image as input of network. In the software environment of MATLAB6.0, this method was proved effective.2. The paper given a integrity knowledge for the image collection system in Hardware-Software and the procedure for the image process. All the date used in vision system is coming from image, so a good and proper image algorithm will saving much compute time and enhance the robustness of the system. Based on this fact, the paper summarized the commonly used algorithm which include image preprocessing, image segmentation and edge detection. This gives a reference for the path planning.3. Proposed a new path planning method based on Genetic Algorithm(GA) for intelligent wheelchair. As a difficult problem for many years, the research and development of it had always been concerned by people. The essence of path planning is a process to finding the best way through iterative and optimization. For the GA is good at iterative and optimization, it is easy to connect the path planning and GA. In this paper, several simple mode was built to simulated environment, then using the VC++ programme to implement the process of GA, and find the feasible way. The experiment showed that the method is effective.The study mode and object in this paper is for intelligent wheelchair's stereo vision navigation, however, it also can be applicated for other motive robot. |
PDF Full Text Request