| Summary: | 碩士 === 國立中興大學 === 電機工程學系所 === 106 === In the application of machine vision, the process of taking the position error or image feature error as the control error, the robot arm or the mobile robot from the initial position to the target position is called the visual servo. In most methods, the control law of the visual servo encounters practical problems such as the occurrence of the local minimum, the unpredictability of the tracing path, or the difficulty of detecting the target feature. In this paper, the feature detection algorithm is used to identify the target object from the current image, and then use the image coordinate position of the target object to switch from the mobile robot search stage, and then use the principle of the depth value visual servo to carry on the current image and the target image The minimization of the convergence of the action, so that the robot can be expected from the initial location to reach the desired target location, and then achieve the purpose of path planning. In this paper, we use the speed up robust feature (SURF) to detect the feature points, and use the random sampling screening algorithm (RANSAC) to further filter the feature points to optimize the feature detection. In this paper, we design a set of two-stage visual servo system with error judgment mechanism, so that the mobile robot can successfully find the object through the rotation search method in the initial situation and the initial image and target the target location by object image coordinates in order to strengthen the depth of the visual servo success rate, and thus to strengthen the previous method of deep vision servo system practicality. Finally, through the experiment, it is proved that the method is applied to the feasibility of path planning.
|
|---|
