| As the plastic industry continues to grow at a fast pace,the pollution caused by bag type plastic garbage has become increasingly evident,which have had a significant negative impact on the ecological environment and seriously jeopardized people’s survival and safety.At present,the cost of degradation of bag type plastic garbage is high and the recycling rate in garbage treatment enterprises is very low,so it is often disposed by landfill and incineration.The speedy advancement of intelligent technology has also expedited the industrial progression towards cost reduction and efficiency enhancement.In addition,smart grabbing has emerged as the predominant method in the field of sorting.However,there is currently no mature smart product or solutions available for processing bag type plastic garbage.Based on this,this paper carries out related research on the identification and grabbing of bag type plastic garbage mixed in household garbage,so as to automatically grab bag type plastic garbage from mixed garbage and reduce its pollution to the ecological environment.Specific work arrangements are as follows:First of all,in view of the problems of difficult detection and easy error detection in mixed bag type plastic garbage identification in household garbage,this paper designed and introduced new network modules C3AM and ASPFF to improve the poor performance of YOLOv5 under the condition of changeable target morphology,perspective characteristics,partial occlusion of target and complex recognition environment.At the same time,by adopting the algorithm of Contrast Limited Adaptive Histogram Equalization(CLAHE),the image’s detailed information has been enhanced and the recognition performance of bag type plastic garbage with unclear boundaries has been improved.The experimental results demonstrate that the improved algorithm presented in the article outperforms the original algorithm,achieving an average accuracy improvement of 2.3%.Meanwhile,the weight parameter is increased by only 0.1MB,which has a good recognition effect in typical scenes.Secondly,the visual system of "eye out of hand" was built.Furthermore,camera calibration was performed to obtain the intrinsic parameters and distortion coefficients of the camera.Subsequently,a "nine point calibration method" was employed to calibrate the camera and the gripping device,resulting in the transformation relationship between the camera coordinate system and the gripping device coordinate system.Thirdly,D-H method is used to establish the connecting rod coordinate system of the grasping device,and further establish the kinematics and dynamics model of the grasping device.After that,the construction of the grasping device are completed.In 10 groups of visual guided grasping device operation experiments,the mechanical claws of the grasping device moved to the preset "cross box under the guidance of the visual system.And The correctness of the visual calibration results,gripping device kinematic and dynamic models were verified through experiments.Finally,a comprehensive experimental study was carried out.The bag type plastic garbage grabbing system loads the pre-trained model and then uses the camera to capture real-time scenes,and the location information identified by the bag type plastic garbage is sent to the grasping device through serial communication,driving the grasping device to complete the automatic grasping task of the bag type plastic garbage.In the mixed garbage scenario,the experimental results show that the comprehensive recognition rate of the bag type plastic garbage grasp system reaches 87%,the error detection rate is only 1.3%,the success rate of grasping has reached 83%,and it takes 27 seconds to grab a target,which has certain practical application value. |
PDF Full Text Request