Research On Human-Robot Integration Safety System Based On Vision Guidance

Under the background of our country’s implementation of the strategy of manufacturing a strong country,industrial robots have achieved rapid development in the field of manufacturing,and the human-robot integrated working model will become the development trend of the future manufacturing industry.The first thing that needs to be solved when the human-robot integrated work mode is introduced into the actual production operation scene is the safety problem.The existing industrial robots lack the ability to intelligently perceive the workspace,and cannot obtain the changes of the human body motion state in the workspace in real time,so it is difficult to timely and effectively.Take control.Therefore,this paper focuses on the theme of personnel safety in the humanrobot integration work mode,and takes the accurate acquisition of human motion information and the evaluation of human-robot safety index in the human-robot integration workspace as the research object,and conducts human-robot integration based on vision guidance.The study of safety systems.The main contents of this paper are as follows:Firstly,aiming at the problem of accurate acquisition of human motion information in the human-computer integration workspace,a human motion information estimation algorithm with skeletal joint point constraints is proposed.According to the human body point cloud data collected by three Kinect2.0 cameras,a human body model is established and the joint point information of human bones is extracted.On the basis of the constraints of human bone length and joint angle,combined with the Unscented Kalman Filter(UKF)algorithm Estimate human motion information.It can be seen from the experiments that the proposed algorithm can obtain the motion information of human skeleton joints stably on the basis of effectively improving the calculation accuracy of human skeleton length.Secondly,aiming at the human-robot safety evaluation problem in the human-robot integrated workspace,a human-robot safety index evaluation method based on Extreme Learning Machine(ELM)is proposed.Firstly,the Speed and Separation Monitoring(SSM)method is used to build the human-robot safety index database based on the human motion information and the motion state of the industrial robot.prediction models for comparison.Experiments show that the proposed method can not only overcome the shortcomings of long modeling time and low accuracy of the BP neural network prediction model,but also can directly map the nonlinear relationship between human-robot motion information and human-robot safety index,which can provide the basis for the subsequent human-robot integration.Security systems provide a reliable source of data.Finally,a human-robot integration safety system based on the Robot Operating System(ROS)is constructed,the trajectory planning and control methods of industrial robots are analyzed,and human-robot integration simulation experiments are carried out to realize the acquisition of human point cloud data,human skeleton joint point information extraction and human motion information estimation,human-robot safety index evaluation,industrial robot control based on evaluation results and other functions.The experimental results show that the constructed human-robot integrated safety system achieves the expected effect,and can ensure the safety of personnel in the human-robot integrated working mode.