Research on Moving Arm Grasping Based on Computer Vision

Abstract

mechanical arms are being used more and more frequently to grab objects in real life like smart-garbage-removal robots. Aiming at the problem that the traditional mechanical arm grasping method lacks intelligence and the eye in hand manipulator is less researched, this paper proposes a mechanical arm grasping system based on computer vision, which uses Yolo object detection to narrow the detection range, reduce noise, use Canny edge detection and image expansion to corrode the center point of the extract body, use Hough to transform the slope of the main direction of the extract body, calculate the position of the object according to the center point and the main direction slope of the object, then use the ROS framework to control the robot arm for grasping. After experimental testing, this method can effectively grasp objects within a certain range, which is more concise than traditional methods, and has good real-time performance and intelligence.