Apple Picking Robot Target Recognition And Trajectory Planning

In the current environment of social development,the agricultural sector has undergone earth-shaking changes.It is developing in the direction of scale,automation,and precision.On the one hand,it is to promote social progress,and on the other hand,to solve the problem of severe population aging.Problems such as insufficient labor and rising labor costs.In the process of agricultural operations,the picking of agricultural products has always been a time-consuming and labor-intensive process,so the research of picking robots has become a realistic need.This paper takes the apple with the highest yield in China as the object,and studies the identification and location of mature apples in the natural environment and the problem of manipulator trajectory planning.In the natural environment,there are some differences between the mature apple area and the background area,but it is seriously interfered by the occlusion and shadow of the branches.In this paper,the gray histogram of each area in the image is studied,and the chromatic aberration coefficient is determined through a large number of experiments.The color feature segmentation and K-means cluster segmentation effects are compared.It is found that the segmentation effect is better than K-means cluster segmentation when the color difference coefficient is 0.8.Finally,color feature segmentation combined with canny edge detection is used to extract the apple outline.For the contours extracted from the non-overlapping fruit images,the minimum circle wrapping method is used to obtain the center coordinates and radius of apples.For overlapping fruits,they need to be divided into individual apple regions by using the pit-based overlapping segmentation method.The minimum circle wrapping method extracts the center and radius information.The method of minimum circle wrapping can extract the center and radius information more accurately when the fruit is partially occluded,which reduces the requirement of image segmentation.The binocular vision system was built,and the left and right camera calibration experiments were performed to obtain the internal parameters and external parameters of the two cameras as well as the distortion coefficients,and the distortion correction was performed according to the distortion coefficient.The binocular camera is calibrated to obtain a rotation matrix and a translation vector describing the left and right camera spaces,in preparation for the three-dimensional positioning of the obj ect.According to the DH modeling method,the mathematical model of the five degree of freedom robot is established and the forward and inverse kinematics analysis is carried out.The space trajectory planning of the mechanical arm joint is performed according to the fifth-order polynomial interpolation method,and mechanical arm model is established in Matlab Robotics Toolbox.The end of the arm is simulated and the velocity and acceleration curves are obtained to verify the rationality of the traj ectory planning.