Trajectory Correction in Robot Moving Path Based on Computer Vision

• Main Authors: Yi-ChengSu, 蘇奕丞
• Other Authors: Tzone-I Wang
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/mvty4y

碩士 === 國立成功大學 === 工程科學系 === 106 === There are a lot of buildings in a city. Robots is helpful for human if they have the ability to move across floors. Elevator is a common equipment in an office building. Therefore, one solution to move across floors is to manipulate and take the elevator. In addition, after a long-distance movement, the error of the position and the orientation of the robot will be accumulated. In order to keep moving in a building stably and not to lose its position and orientation, a robot must have the ability to localize itself. Camera has been almost a necessary sensing equipment for various modern robots. If a robot can remember the surroundings and the passable path, make decisions and localize itself just like human with vision, much redundant expense of sensors can be reduced. It will make intelligent robots more available and affordable. In this study, the robot, equipped with a monocular camera, a robotic arm and wheels, is developed to complete delivering tasks across floors in a building. The error, accumulated while the robot is moving in a floor, can be reduced by calculating the ego-position via image processing technique on acquired images. Correcting the trajectory of robot can provide stabilized movement to destination. To manipulate the elevator autonomously for crossing floors, the robotic arm is drove after the computer recognizes the button on the panel and calculates its position. A mobile application is developed as user interface of the robot to make it easy to be used by user. Entering destinations on the app triggers the main program of the robot. The robot will be back to the starting point after finishing delivering tasks.