| At present,the collection of goose breeding eggs in China is mainly manual work,with high labor intensity and low collection efficiency.During the breeding process,workers frequently enter and leave the goose house to work,which could easily cause stress reactions in breeding geese,affect egg production efficiency,and increase the risk of human and animal cross infection.In order to solve the above problems,based on existing domestic and foreign egg collection robots,this paper designed and developed an intelligent collection robot for goose breeding eggs in unstructured environments,which meets the requirements of mechanization,intelligence,and unmanned collection and transportation of goose breeding eggs in unstructured environments.At the same time,it provided a theoretical and device basis for the collection of poultry eggs in unstructured environments.The main research work of this article was as follows:(1)The overall structure and functional requirements of mobile platform of the intelligent goose breeding egg collection robot were determined through the analysis of the goose breeding house environment and the law of egg production behavior of the breeding geese.Aiming at the hard road environment in the goose breeding house,the chassis structure of the mobile platform was designed,and through the simulation analysis of its inherent suspension characteristics,suspension parameters were obtained to ensure the trafficability of the mobile platform during driving.A navigation and obstacle avoidance system for the mobile platform has been developed for obstacles such as barbed wire,red bricks,geese and feeding pipes in the goose breeding house,ensuring the safety of the intelligent goose breeding egg collection robot.In addition,in order to meet the grasping requirements of different positions in the egglaying area,a 3-degree of freedom manipulator structure was determined through simulation of the operating range of the manipulator.(2)Aiming at the problems that the color of the padding on the egg bodies was similar to the background color in the non-cage environment,and the goose eggs were easy to be buried in the padding by the breeding geese which resulted in incomplete recognition,a goose breeding egg recognition algorithm in the non-cage environment was designed:Firstly,the goose breeding eggs in the captured image were detected by YOLOv5,and the original images were segmented according to the confidence coefficient to obtain goose breeding egg segmentation images;By processing the background of the segmented goose breeding egg images,the goose breeding egg contour pixels were obtained,and the goose breeding egg contour curves were generated by combining the goose breeding egg contour curve construction equation;Then,the segmented images were put into the deep learning module for training.Combining vector stitching feature fusion technology and support vector machine(SVM)feature classification technology,the goose breeding egg recognition of the segmented images was realized;Finally,based on the centroid positioning and coordinate system conversion equation,the pixel coordinate information of the goose breeding egg centroid is obtained.In addition,comparative experiments were designed to verify the effectiveness and superiority of the proposed algorithm.(3)Using the pixel coordinates of the goose breeding egg centroid as input,the coordinate information of the goose breeding egg centroid was determined combined with the binocular camera positioning principle.Aiming at the shortcomings of current robot path planning algorithms that only considered the optimal motion time of the robot and ignored the optimal path selection,a path planning algorithm that simultaneously searches for the optimal path node and time node was designed:Firstly,an improved D-H method was used to construct the robot motion model;Then,the motion path of the manipulator was fitted using a cubic Bspline curve;Finally,a fast non dominant genetic algorithm with elite retention strategy(NSGA-Ⅱ)was used to simultaneously search the path nodes and time nodes of the robot to find the optimal path curve.In addition,comparative experiments verified the rationality and superiority of the proposed algorithm.(4)The mechanical structure and algorithm program of the intelligent goose breeding egg collecting robot were integrated.By analyzing the motion characteristics of the motion devices,the hardware parameters of the motion device were determined,and the hardware selection was completed.Aiming at the stress response of breeding geese,through the operational logic design of the intelligent goose breeding egg collection robot,the operational sequence between various hardware modules was planned,the communication protocol between various modules was clarified,and the logical design of the platform movement control module,visual positioning control module,joint rotation control module,and end effector control module was carried out,achieving coordinated control between various modules.(5)The intelligent goose breeding egg collection robot was developed.A dark room environment was built in the egg-laying area.The success collection rate of the intelligent goose breeding egg collection robot under different operating times,different lighting angles,and different lighting intensities was tested.The experimental results showed that the success collection rate of intelligent goose breeding egg collection robot in daytime room was 93.3%;The success collection rate at night was 83.3%.In addition,in a dark environment,when the illumination angle of the supplementary light source was 60° and the illumination intensity was between 500 lux and 600 lux,the maximum success collection rate of the intelligent goose breeding egg collection robot was 86.7%,which is close to the effect of daytime collection. |
