Research On Fruit-harvesting Robot Foundation Platform Combines With WLAN

China is the largest producer of fruits in the world. Currently, the fruit-harvesting is carried out by hand without any effective equipment. With the trend of aging society in China, the fruit-harvesting is facing the problem of lack of labor. Therefore, researching and realizing the fruit-harvesting robot for market have its practical significance.For the problems above, conduct works include:(1)There are lots of research results about autonomous fruit harvesting robot at home and abroad, but it is still unable to meet the market requirements. To solve this plight, a human-computer remote collaboration platform is achieved with combining the WLAN and robotics. It gives fully play to object recognition by people and execution by robot, thereby improve the efficiency of the fruit-harvesting.(2) Fruit-harvesting is arms tight coordination task in dual-arm robot platform which requires high accuracy positional and speed output by motor drive unit. To suit this requirement, the distributed motion control system based on EtherCAT protocol is deployment. Tests show that the servo system achieves positional dynamic bias within ±3 counts, static error within ±1 counts and the speed dynamic bias within ±%5 in low-speed level.(3) Human-machine interaction is an important part of remote collaboration platform. Design and implementation an interaction method based on the depth image which supports specify the three-dimensional space target in a two-dimensional interface in a more natural way. Another three interactive panels were designed to support system functions under Windows systems. They are duties, clearly defined and task logic. Interface is running smoothly and interactive information presentation clearly.(4) The image segmentation for target object is the key technology of fruit picking robots. An adaptive color threshold segmentation method was employed for accurate greenhouse-tomato image segmentation from background. The OTSU segmentation method was improved making it suitable for this issue.