Research On The Pneumatic Harvesting Robot Based On Machine Vision

With the development of technology, automation production is the inevitable trend, agricultural production is no exception, and China government has already identified the strategy of developing the country by rely in science and education. More automation equipments need to be applied in agricultural production to increase productivity, increasing productivity and reducing labor costs is the future direction of agricultural development. Fruit-growing industries occupies a pivotal position in agricultural production, but fruit picking almost all depends on human labor in our country. With the development of social economy and accelerating of ageing populations, agricultural labor shortages will be a big problem, the fruits production cost will also increase, So to realize the fruits picking automation makes a great significance for the agricultural development. Currently agriculture harvesting robot is still in the research stage, there is still a long way to go before realize practical and commercial, so develop a high-efficiency, low-cost agriculture harvesting robot has a long-term significance.A robotic decice consisting of a manipulator with 4 DOF, binocular stereo vision system and pneumatic proportional control systems is developed for apple harvesting. A series of research work has been done for this robot, the main research work is presented as follow.Firstly, according to the characteristics of the agricultural harvest and the actual requirements, this paper designed a harvesting robot use pneumatic transmission as the basic drive mechanism, the 3D models of the manipulator with 4 DOF has been developed by UG sofware, then used Auto CAD software to draw mechanical processing of two dimension drawings. Besides, the pnuematic drive system for robot has been designed,complete the task of pneumatic hardware selection.Secondly, the link coordinate for agricultural harvesting robot is established by D-H method, on this basis kinematic modeling and analysis for the robot has been done, then get the forward and reverse kinematics results. Matlab software are used to planning the path of robot in joint space, after the motion simulation has been done in UG software, the position forward and reverse solution has been proved correct.Thirdly, a stereo matching image recognition algorithm based on apple centroid characteristics is proposed and the coordinates of apple centroid is obtained using triangulation principle, and the stereo vision program are designed using Visual Studio 2010 software.Fourthly, dynamitic of manipulator and pneumatic system modeling has been done, then the simulation model has been setup by AMESim software. The strategy of common PID and fuzzy adaptive PID has been proposed, the co-simulation between AMESim and Simulink also been done.At last, built the control system for pneumatic harvesting robot, and completed the work of relevant hardware selection. The control program is designed using MFC and the prototype of pneumatic harvesting robot is built. Then corresponding experiments has been done on the basis of the prototype, and analyzes its overall performance and problems.The experimental results show that the positioning accuracy and repeatability of pneumatic harvesting robot can meet the accuracy requirements of agricultural harvesting. And its binocular stereo vision system can identify and locate the fruit automatically, its control system can guide the robot complete the auto-picking task.