The Research On Obstacle Avoidance And Motion Control For Intelligent Wheelchair Based On Laser Sensor And Monocular Vision

With increasing attention of the society for the aging population and disabled group, the demand of wheelchair is augmenting which have promoted its level of intelligent and autonomous and attracted the attention of researchers. In view of the application background of intelligent wheelchair, the security and stability should be considered preferentially in the study, and the design of avoidance strategy is disparate according to the characteristics of different intelligent wheelchair platform.In this thesis, the obstacle avoidance method for intelligent wheelchair is deeply studied and the motion control system is optimized based on the platform which is our independent research and development. Firstly, aiming at the shortcomings of the laser sensor for obtaining less environmental information and single level, the obstacle avoidance strategy is introduced based on monocular vision. Then for controlling wheelchair speed more fluently, the fuzzy PID control strategy is proposed. It can significantly enhance the running stability and timeliness.The main research work in this thesis as follows:1) Obstacle detection method based on monocular visionThere are many background image constraint conditions for detecting object based on background subtraction method. To solve these problems, this paper presents obstacle detection method based on fusion background subtraction. Use background subtraction method to achieve the obstacle detection, and use frame difference method to replacement background image.This method can effectively eliminate the interference of light and other factors. That can achieve precise obstacle detection.2) Multi-sensor information fusion obstacle avoidance based on laser and monocular vision sensor.In order to compensate the lack of lower obstacle information by laser, we present vision sensor improve the environment information. This method realizes laser and monocular information fusion based on Bayesian theory. Enhance the perception and comprehension of the system for obstacle information in real environment. This algorithm improves the limitation of single sensor in process of obstacle avoidance, further improves the robustness of system.3) Omnidirectional intelligent wheelchair motion control based on fuzzy PID method.The wheelchair obstacle avoidance system demands high response speed. In view of this, we proposed speed control system based on fusion expert system and fuzzy PID. It builds the knowledge base and the fuzzy control rules by expert experience. It controls the wheelchair speed and direction through user expects motion result. Input the decision result of expert system and actual velocity of the wheelchair into fuzzy PID controller. Then use the designed fuzzy control rules to online self-tuning P, I, D parameters. Final, achieve real-time and precise seed control. Also can improve the system dynamic performance and lower the system overshoot.Experimental results show that the proposed information fusion obstacle avoidance method based on laser and monocular can detect obstacle information more effectively and robustly. The fuzzy PID controller can accurately implement the avoidance for obstacles. And the system response speed and accuracy are within a reasonable range.The research in this thesis has a certain theoretical significance and practical value in the intelligent wheelchair autonomous navigation and obstacle avoidance.