Research And Realization Of Intelligent Wheelchair Localization Based On Vision

Intelligent wheelchair is developed based on the technology of mobile robot, integrating machine vision, pattern recognition, automatic control, multi-sensor information fusion and multimodal human-computer interaction and many other high-tech as a whole, with automatic positioning, navigation and obstacle avoidance function, can provide help for the disabled people, improve their quality of life. Along with the increasing of the aging of the population and the disabled, the society will be more urgent needs intelligent wheelchair. So it has very important realistic meanings to study the intelligent wheelchair.Localization is the key technology to realize the autonomous navigation of intelligent wheelchair, is also the key and difficult problems in this research field. This paper designs an intelligent wheelchair platform, and then this paper discusses the vision-based localization problem of intelligent wheelchair on this platform.This paper mainly does the following work:(1) Based on the study of the intelligent wheelchair’s research situation at home and abroad as well as the main technology of the intelligent wheelchair, we design a intelligent wheelchair platform which localization based on machine vision and obstacle avoidance based on ultrasonic sensors, the platform adopts portable PC and Arduino Mega2560development board for the control structure. The architecture of the platform is open, additional functionality and modules can be added easily in the future.(2) The theory of location based on vision of intelligent wheelchair is studied. From the view of pattern classification, global location for intelligent wheelchair can be seen as a scene recognition problem, and then propose a method based on principal component analysis and support vector machine’s scene recognition algorithm for location. Scene images are processed by the algorithm of image preprocessing firstly, and then use the principal component analysis for feature extraction to get the global features of the image, and then use the support vector machine algorithm to classify. The method proves to be effective through the simulation experiment.(3) Finally, we developed a localization system based on scene recognition algorithm by VC++6.0and OpenCV, so as to realize the self-localization for intelligent wheelchair based on vision.