Feng Qingchun

Study on strawberry robotic harvesting system

In order to improve robotic harvesting for strawberry and reduce production cost, the harvesting robot system was designed for table-top culture, which was able to harvest fruits on both sides of itself, and it was supposed to served for sightseeing agriculture and technological show through integrating wireless operation and voice promotion. Based on machine vision and sonar technology, the independent navigation system for harvesting robot was built. The fruits were distinguished according to H(Hue) and S(Saturation) color histogram and the picking-point was located by binocular-vision system. The nondestructive operations for fruit include sucking fruit, holding and electrically cutting fruit-stem, in order to prevent damage on pericarp and disease infection. And the end-effector was positioned by a joint-type industrial manipulator. Finally, as the experimental results showed, the successful harvesting rate was 86%, and every successful harvesting operation on average cost 31.3 seconds, and the average error for fruit location was less than 4.6mm.

Selecting candidate regions of clustered tomato fruits under complex greenhouse scenes using RGB-D data

This paper presents a new candidate region selecting strategy for detecting clustered tomato fruits using RGB-D data, which is taken in complex greenhouse scenes. The strategy employs both color and depth information in the same Kinect sensing frame. First, one Kinect frame is processed to generate three separate images, including depth image, RGB color image, and S channel image in HSI color space. The images are processed with depth threshold, green enhancement and saturation threshold correspondingly. And a frontground mask image will be generated with the processed images through intersection and union operations. Second, depth gray image and color gray image will be filtered with the frontground mask. And candidate fruit regions can be obtained after open-close reconstruction. Experimental results show that the strategy can greatly cut down the regions of interests while keeping all the true fruit regions.