Research And Application On Intelligent Interaction Technology Of Desktop Robot ARM

After decades of research and development,industrial mechanical arm has made many achievements and been widely used in industrial production.Desktop robot arm is different from industrial mechanical arm working in a closed environment,intelligent interaction of desktop robot arm in the human-machine environment is facing many challenges.Desktop robot arm is mainly used for family entertainment and teaching.Several factors,such as the dynamic and uncertainty of the domestic environment,tens or hundreds of different target objects,arbitrarily placed poses,contacts and occlusions between objects,etc.,impose higher requirements for the intelligent grasping of desktop robot arm.Compared with the popular industrial mechanical arm,the desktop robot arm has the characteristics of small size,low price and more demands for human-computer interaction.In order to better realize the human-machine interaction function of the desktop robot arm,the main research contents and results of this paper are as follows:1)Adopt the modular design and choose Raspberry Pi as the desktop robot arm's core control board.Besides,design and develop the middleware code and robot arm's driver control board.All these steps were designed to ensure that the desktop robot arm can accurately move in the complex man-machine environment.Ultimately,complete the target object of grasping task.2)Add human-machine voice interface for desktop robot arm: as an essential technology for intelligent products,voice recognition has become more and more important in the application of intelligent products.LD3320 speech recognition system has the advantages of flexibility,reliability,small size,low power consumption and high recognition accuracy.In this paper,the LD3320 is used toenable the desktop robot arm to "understand" human language and realize human-machine interaction in the information age by using "voice",the most natural and convenient means.so that the desktop robot arm becomes more valuable.3)Add machine vision for desktop robot arm: At present,there are many application frameworks to realize machine vision.According to the working environment and nature of desktop robot arm,it is another key point of this paper to build an appropriate visual application platform to improve the accuracy and efficiency of machine vision recognition.In this paper,the target detection algorithm was studied.The Faster R-CNN target detection architecture was adopted,and zhang zhengyou calibration method is used to obtain the camera internal parameters and distortion parameters.Combined with the hand-eye matrix,the coordinate position of the target object in the robot arm coordinate system is solved,and the desktop robot arm is driven to complete the task of grabbing the target object.The test results show that the method based on deep learning is more accurate than using OpenCV alone.