Obstacle Avoidacne Of Intelligent Wheelchair Based On Multi–sensors Information Fusion And Motion Control

In order to help the old people and handicapped people, the role of multifunctional intelligent wheelchair becomes important to enhance their mobility. The key technologies of intelligent wheelchair draws more attention from researchers and becomes a research hotspot in the field of mobile robot.To ensure the safety of users, it is demanded that the intelligent wheelchair should have a faultless obstacle avoidance function. This paper focuses on the method of obstacle avoidance based on omnidirectional intelligent wheelchair. First of all, to solve the problem of traditional optical flow algorithm which have to calculate large amount of data, we present a new obstacle avoidance method based on the balance of optical flow algorithm. Furthermore, to improve the limitations of ultrasonic sensor or monocular sensor, we present a multi-sensors information fusion obstacle avoidance method based on ultrasonic sensor and monocular sensor. Finally, we present an omnidirectional intelligent wheelchair motion control system based on fuzzy PID, which improved the obstacle avoidance effect of intelligent wheelchair,The main research works are listed as follows:1) Optical flow balance algorithm of obstacle avoidance based on monocular visionTraditional optical flow algorithm needs to process large amounts of data when computing the obstacles information, and has a poor real-time performance. In this paper, we provide an obstacle avoidance method based on the algorithm of optical flow balance. This method can estimate the field intensity of obstacles. It has a high real-time performance and needs a small amount of obstacle information. It is suitable for intelligent wheelchair obstacle detection and improves the effect of obstacle avoidance.2) The information fusion based on monocular vision and ultrasonic sensor for obstacle avoidanceUltrasonic sensor is popular in mobile robot application due to its simple information processing procedure. It can execute obstacle avoidance by using improved fuzzy neural network algorithm. But the sound wave interference easily happens under the condition of dense obstacles, while monocular vision sensors can obtain obstacle information accurately. This paper presents a new obstacle avoidance method based on the information fusion from monocular vision and ultrasonic sensor. We use the Bayes fusion method to combine the vision and sound wave information, and on this basis to realize omnidirectional intelligent wheelchair autonomous navigation and obstacle avoidance. It avoids the isolated processing for information from single sensor, and also avoids the lack of information when mobile robot explores the unknown environment3) The motion control system of omnidirectional intelligent wheelchair based on the fuzzy PIDAccording to the requirement for obstacle avoidance, an expert system combined with fuzzy PID controller was designed. The expert knowledge and control rule set were established by the user’s intention and experience. Thus the wheelchair speed was determined according to the occupant’s tilting degree. Furthermore, the output of the expert system and current velocity of wheelchair were used as the inputs of fuzzy PID controller. The fuzzy rules were adopted to adjust PID parameters to trimmer the speed precisely. Consequently, the wheelchair velocity control system showed better dynamic performance and lower overshoot.The experimental results showed that the proposed method can improve the obstacle avoidance effect of omnidirectional intelligent wheelchair and have practical application value. The motion control system of omnidirectional intelligent wheelchair based on the fuzzy PID can precisely execute the control instruction coming from the obstacle avoidance system.