Visual Positioning Of Fruit Harvesting Robot In The Situation Of Disturbance

Harvesting robot is introduced in the field of fruit picking in order to adapt to thedevelopment of agriculture, reduce labor cost and improve labor productivity. Thework environment of fruit picking robot is complex, so the development of fruitpicking robot needs to solve many critical problems.The identification andlocalization of ripe fruit are two key problems which need to be solved.Orange harvesting robot was taken as research object, and the recognitionmethod and spatial orientation method of oscillating orange were researched. In theprocess of orange picking,every kind of picking method will inevitability causenearby oranges to irregularly oscillate.The traditional static method of orange pickingis to pick orange until oscillating orange is actionless.The method can not obtain thespatial coordinates of dynamic orange,influence working efficiency of picking robotand cann’t adapt to the picking need of oscillating orange.The main research results ofthis paper are as follows:In order to recognize covered and oscillating orange in the nature scene. Therecognition method of oscillating orange was proposed, which was based on framedifference method.Firstly, I2component was used to separate orange from background,and iterative segmentation method was used to segment orange image.Secondly, theinter-frame difference method, the minimum enclosing rectangle method of horizontaldirection and the square inscribed circle method were used to labeled the dynamicregion of oscillation orange.The oscillation orange was located and identified throughthe centre of a circle.Spatial orientation method of oscillating orange was researched Under-dampedoscillation model was used to simulate orange oscillation process, FFT spectrumanalysis method was used to determine the oscillation period of oscillation orange,and Fourier curve fitting method was used to determine the centroid coordinates ofequilibrium point.Three-dimensional coordinates of equilibrium point was calculated based on the centroid coordinates of left orange image and corresponding right orangeimage.According to oscillation period and three-dimensional spatial coordinates ofbalance point, the trajectory of oscillation orange was determined and automaticpicking of oscillating orange was realized.