Study On Key Technologies Of Loquat Pruning Robot

The pruning of fruit trees is an important technology to improve the fruit quality, enhance the disease resistance and reduce the frequency of on-year and off-year. At present, the pruning of fruit trees is still dominated by manual pruning. However, with the transfer of labor force and the aggravation of the aging in rural areas, the low efficiency of manual pruning method is no longer suitable for the development of fruit industry. Thus, this paper, taking loquat branches as the object of the research and combining the technology of agricultural cultivation, industrial robot and branch recognition, conducts research on mechanized pruning. The process of mechanical pruning mainly includes branches recognition, camera calibration, determination and location of pruning branches, the pruning process and so on. Among all the processes, the related research of branches recognition, camera calibration as well as the pruning process are far from mature stage. Therefore, this paper mainly carries out in-depth research on these three aspects and the specific contents for research are as follows:It dissects traditional planting method and production system of loquat, defines loquat’s planting system, analyzes the possibility and position of mechanization of the loquat planting, and establishes the loquat pruning robot’s function in loquat planting system. It determines the key technology of loquat pruning robot as branch image recognition, the ultra-wide angle camera calibration and the development of the pruning end-effector according to the technical analysis of loquat pruning robot. It makes a comprehensive analysis of the pruning operation environment and also designs an overall scheme of the loquat pruning robot on the basis of achieving pruning with the combination of mobile platform and cylindrical coordinate robot. It also obtains the motion parameters of the loquat pruning robot as well as the natural vibration frequency for different position with the method of kinematics analysis and modal analysis of the loquat pruning robot. In addition, it verifies the feasibility and the value in engineering application of the overall scheme of the loquat pruning robot by means of the numerical simulation of the parts and whole loquat pruning robot.It carries out the related research into loquat branches recognition, For example, it undertakes exploration of the segmentation of loquat branches’image, noise elimination, the compensated image processing of unreasonable intermittent, the extraction of the framework of the image, the acquisition of the branch diameter and so on, and proposes two algorithms with different features for loquat branches image recognition and frame extraction, and the experimental study was carried out. Experimental results show that these two algorithms can obtain better recognition results than OTSU adaptive image recognition, and have different advantages in processing speed and correct recognition rate. Therefore, it can meet the requirements of different branch identifications.In this paper, in order to obtain large eye-shot of image information, it collects the branches image by the use of an ultra-wide angle camera and according to the principle of ultra-wide angle lens’s distortion, an ultra-wide angle camera calibration algorithm is proposed. The algorithm firstly uses the relationship of the different points in the camera coordinate system and the characteristics of the image distortion to acquire the ideal image coordinates without distortion. And then, according to the constraint relationship between the ideal images coordinates without distortion and the actual image coordinates, the non-distortion ideal image points are corrected. In the end, the inside and outside parameters of the ultra-wide angle camera are obtained by using the corrected image coordinates and the actual image coordinates. By comparing with other algorithms, the algorithm can obtain the camera parameters accurately, and has the advantages of simple operation and fast calibration.Based on the structural characteristics of loquat branches, a method for determining and locating the cutting branches of loquat is put forward. In this method, the diameter, number and the area density of branches are obtained based on the position of binocular camera and the central axis coordinates of branches. Then, the pruning branches are determined based on the diameter and density of branches, taking the intersection of the branches central axis and the pruning line as the cutting point.Eventually, the three-dimensional coordinates of the cutting points are solved by the equation. Experimental study shows that the pruning branches obtained with this method are in accordance with the requirements of pruning, the three-dimensional coordinates of the cutting points are relatively accurate, and it can be used in the pruning operation of the pruning robot.On the basis of the special requirements in the pruning process as well as the load limit of the robot’s arms, it fixs the function of the loquat pruning end actuator, designs and manufactures a pruning end actuator for loquat, and the transmission analysis is carried out. It also acquires the theoretical parameters of the pruning end actuator by the analysis of chucking force, the conveying force and the feeding force. What’s more, a cutting experiment is conducted and the optimal parameters combination for the pruning end actuator is determined on the basis of these parameters.Finally, it carries out the pruning experiment with the help of the key technology such as the loquat pruning robot’s simulation device, integrated camera calibration, branches image recognition, and the end execution pruning device, successfully completed the loquat pruning operation, and confirms the practical value of related technology in engineering.