Research Of Control System Of Robot Grabbing Motion Object

With the continuous development of China’s robot industry,robots play an increasingly significant role in our daily life.Grab robots are widely used in industries,agriculture,logistics and services.Grabbing robots on industrial production lines can reduce the workload of workers and improve the productivity and quality of products.In addition,in some polluted areas such as high temperature,noise and dust,as well as in narrow working space,the grabbing robot can directly replace human to complete the work and protect human life and safety.The home service grabbing robot can help the elderly to grab books,water cups and other daily necessities,and improve the quality of life.In this paper,by predicting the trajectory of a moving object,a grabbing robot can grab the target in time and space.This paper presents a prediction algorithm for trajectory recognition of moving objects,designs a control system for the robot to grab moving objects,and completes the design of corresponding hardware and software systems.The research process is as follows: first,data collection and preprocessing,motion track recognition and prediction,then the inverse kinematics solution of the robot is studied.Finally,the position coordinates of the current test track are predicted through the motion object track collection system,the end of the robot grip is controlled to move to the predicted position,and the capture of the moving object is completed,which verifies the feasibility of the system.This paper studies from the following aspects:(1)Designed a trajectory recognition and prediction algorithm for moving objects.First,the original data collected is preprocessed,including data correction,data fusion and track filtering.Then,the random track is divided into basic tracks by using the special point segmentation method,and the image slope algorithm is used to convert the random track into a mirror slope sequence.The basic tracks are clustered again,the basic track library is established,and the basic tracks to be measured are identified.Finally,LSTM network is established for each type of track in the basic trajectory library to complete the trajectory prediction.(2)A mirror slope algorithm is presented.Because the change of oblique angle cannot directly reflect the change of oblique angle slope and the range of slope change is too large,a mirror slope algorithm is proposed to convert the segmented basic trajectory from the slope sequence to the mirror slope sequence,so that there is a corresponding relationship between the oblique angle and slope,and the range of change is smaller.(3)ACC-ULS-TP-CLUS clustering algorithm is presented.To solve the problem that traditional clustering algorithms are difficult to achieve direct clustering of nonequal length time series,we first propose the ACC-ULS-TP algorithm to compute the center of non-equal length time series set,and then propose the ACC-ULS-TP-CLUS clustering algorithm to cluster the ordered columns in non-equal length time series set.The ACC-ULS-TP-CLUS algorithm is compared with FE-SOM-BPNN and FEKMEANS methods in building the basic trajectory library and identifying the trajectory to be measured.(4)A prototype of the grabbing robot is built.Firstly,the whole structure of the grabbing robot is designed.The robot prototype mainly consists of four rotating joints,five connecting rods and one grip.The overall structure of the robot is modeled using the improved D-H parameter method,and its forward and inverse kinematics are solved.(5)A control system for robot grabbing moving objects is designed,which is composed of two parts: a track collection system for moving objects and a control system for robot grabbing.Among them,the motion object track collection system consists of data collection node,master controller and data sending node.It mainly completes the motion object track collection,track recognition and prediction,and sends the prediction results to the robot capture control side.Robot grabbing control system consists of data receiving node,master controller and grabbing robot.When the predicted result is received,the master controller can control the robot to move to the target position and achieve grabbing by coordinate transformation and inverse kinematics solution.