Control Technology Of Robotic Arm Based On The Ground Combat Platform

With the advent of artificial intelligence technology,the development of intelligent robots has been promoted in battlefield.Robots still have a lot of research space for autonomous grabbing.The current technical difficulty lies in the identification and positioning of target objects.In order to solve target object recognition,this paper develops a neural network recognition system based on ResNet.In order to solve the positioning problem of target objects,a positioning scheme based on binocular camera without external sensors is designed.The research of this paper is carried out through the identification and positioning of different objects,and finally the PC controls the robotic arm to complete the grabbing task through serial communication.The main work of this article mainly includes the following three points:(1)Target identification.Using TensorFlow to build ResNet neural networks,and training our own data sets to build the recognition system.When the target is identified,the left camera in the binocular camera is opened for collecting images and then these images are input to the recognition system for identification.(2)Binocular camera calibration and distortion correction.In order to eliminate the distortion of the camera,the camera is calibrated by Zhang Zhengyou calibration method,and then programming by OpenCV to correct the camera.At the same time,the parameters calculated in the calibration can be used for binocular camera ranging.(3)Grabbing target.Controlling the movement of the end effector of the robotic arm to the target object requires the inverse kinematics of the robotic arm to grasp the target object.Through experiments,we find that the proposed solution can meet the requirements of grasping indoors.During the impact of the day by the light and the object material factors,thesuccess rate of grasping is 80%.The success rate of grabbing is 70%-83% at night.The innovations in this article are:Designing independently the robotic arm grasping scheme that does not require external sensors.It requires only a binocular camera to accurately identify and grasp the target object.