The Research Of Hospital Ward Inspection Robot Localization And Obstacles Avoid

Patient care is a basic task of hospital nursing. Because the monitoring task labor intensity is too large. Clinically, the problem prevalently exists which nurses cannot timely take care of the patients. With the rapid development of the technology of service robot, robots in the medical, welfare facilities and families play an increasingly important role. It has become possible which the robot replaces nurses to complete parts of the patrol work. In order to fulfill the tasks above on, the ward inspection robot should solves the robot’s navigation and obstacle avoidance problems. This paper carried out the following work that based on the hospital environment to research the technology which used for the ward inspection robot to go from the nurse station to the ward accurately, reliably and do the barrier-avoiding actions timely.In view of the hospital environment, this paper presented a new environmental learning and modeling method with the RFID sensor. With the combination of Stargazer and RFID, topology nodes those contain context information are designed in the hospital environment to guide the ward inspection robot to finish the environment autonomous learning, automatically save the key node in the feature information, and construct the environment model. In order to ensure the ward inspection robot to reach the target ward accurately and reliably, this paper used the D-S evidence reasoning in data fusion method with RFID, odometer, vision and Stargazer. It greatly improved the robot’s precise positioning in the large-scale workplace. Between the two topological nodes, the ward inspection robot uses laser detection corridor wall, and linear fitting algorithm to calculate the angle between the robot direction and corridor wall. Using the angle to correct odometer value, the ward inspection robot realizes the precise positioning between topological nodes.Aiming at the complex environment, the ward inspection robot obstacle avoidance method was researched. This paper summarizes the defect of traditional method used in robot obstacle avoidance, and proposes a new method which can comprehensively detect and avoid obstacle by using the laser sensor and camera. The paper also proposes the background subtraction method for ground obstacles and uses threshold segmentation to identify obstacles. Through coordinate transformation, this paper changes the image’s pixel coordinates to the camera coordinates. Using the camera coordinate to calculate the distance between the robot and the obstacles. In order to eliminate the perspective effect on visual location accuracy during the image processing this paper used the bird’s-eye view method and enhanced visual location precision. According to the different obstacles, this paper uses laser and visual to detect and avoid obstacles by combining them, queries the obstacle avoidance strategy table, and then achieves accurate and reliable inspection robot obstacle avoidance action...