Research On Target Recognition And Grasp Control Of The Manipulator Mounted Intelligent Wheeled Vehicle

It is practical and meaningful on studying intelligent wheeled vehicle, especially in present world of informatization and automation. The vehicle, armed by aims-reaching manipulator, has accurate target identification. In this paper, through the camera image acquisition on mechanical arm, both image processing and object recognition are studied, in order to control the robot to grasp target.In recent years, the rapid development of intelligent control has being widely used in automation, electronics, medical and other industries. In the application, the manilulator equipped intelligent vehicle with visual ability does not only apply to the exploration, search and rescue in hazardous areas, but also can be used in the medical industry, for caring the elderly and persons with disability. Theoretically speaking, intelligent robots mostly refers to two aspects:one is to automatically sense the surrounding environment through a variety of sensors. The second is based on the understanding of the surrounding environment, it will automatically take the next step. In this paper,the intelligent vehicle, which equipped with robot system, is the subject of the results of the use of pattern recognition, to achieve cooperation and coordination with the body robot, complete the grasp task. Image processing and pattern recognition are coordinated to complete tasks of the recognizing and grasping of the target with theoretical research and innovation.This subject achieved path tracking, infrared obstacle avoidance, voice broadcast, intel ligent elevation and infrared remote control on an intelligent wheeled vehicle equipped with a laptop, a camera and a 5-DOF manipulator. Laptop is used as the main platform for targe t recognition. Based on the images collected by the camera, using the OpenCV library for i mage pre-processing. Using the "Second Line Identification" feature extraction to plug the r esulting 5x5 feature matrix into the ART neural network learning algorithms for target reco gnition, and determine whether it is the correct target for grasp. The manipulator is controlled by MCU and its motion trajectory is achieved by Jacobian matrix and applying forward and inverse kinematics analyzing to the D-H model. The manipulator has 5 DOFS (degrees of freedom) and each DOF is controlled by a single motor in order to realizing grasping, releasing and lifting. The control unit receives parameters of the target and grasp it accurately. Experiments are conducted and the results are within expectation of the intelligent wheeled vehicle recognizing and grasping targets whiling running...