Study On Caged Layer Health Behavior Monitoring Robot System

At present, laying hens equipment and environmental control technology had made progress in China. However, artificial monitoring was a main method to distinguish the health behavior of caged poultry, which resulted in low working efficiency, physical energy consuming and the poor health for people. As a matter of fact, the sick or died hens not identified effectively were bad for birdhouse epidemic prevention and chicken health, which would restrict the development of laying hens breeding industry. So the mechanization and automation of monitoring layer health behavior was a problem to be resolved to the laying hens industry. Many investigators had been studying the problem by means of machine vision technology because of the special characteristic and complexity of laying hens. However, machine vision technology applied to the layer health behavior monitoring robot had not been studied. In order to resolve the problem, the key technologies for caged layer health behavior monitoring robot were studied using stepwise-feeding hens as object. And the research findings were as follows:(1)On the basis of analysis of traditional laying hens breeding system in China, a breeding system suitable for monitoring laying hens health behavior was provided.(2)A layer health behavior monitoring robot with 3 degree of freedom was designed and manufactured independently according to the basis of analysis of the robots’ mechanical structure which includes basic structure, dynamic type select and driving devices select, which could motion agilely and position accurately. The mobile robot was turned by wheels and linkage device. Executing devices were drived by stepping motors and connected screw units. This robot was equipped with an end-effector and could proceed 3D operating. The stable mechanical structure could be applied to different cultivation system.(3) The hardware and software platform of robotic visual system was carried out. A algorithm based b* component of L*a*b* color model was proposed and used for segment processing of laying hens images based on theoretical research of the underlying image processing. Technology of histogram equalization and 3×3 median filtering method were introduced to improve the images quality. Combining different processing methods such as OTSU method, dilation and erosion, the target segmentation and noise elimination of image of laying hens were processed. And defining interested area and connected area calibration were used to get the number of target objects. And the visual recognition algorithm for experimental results showed there was 90% of success.(4) The hardware and software platform of robotic control system were carried out. The hardware platform included the main controller based on PC and the secondary controller based on PIC16F877A SCM. The applied control program was developed on the hardware platform of control system. Centered by the main controller, the human-computer interaction interface based on windows system was designed to complete the image collecting, image processing and the judgment of target location, and also real-time communicating with the secondary controller. The secondary controller was used to control the movement of every axle of the robot. The robotic kinematics analysis was proposed, robotic kinematics equation of image sensor was set up based on D-H parametric method.(5) The control accuracy of mechanical structure and visual recognition system performance of the robot was tested. The testing including positional accuracy and recognition rate of robot was designed to verify the accuracy of target recognition algorithm and location Algorithm. The test result showed that robotic positioning accuracy and repeat precision of the whole system was not more than 16 mm, recognition rate reached to 87.5%, which indicated the robotic control system was stable enough. Visual recognition algorithm researched in this article could effectively transplant to the robot, which made the layer health behavior monitoring robot have a large application space.