Método de reposicionamento de robôs por entropia de imagem bioinspirado na sensibilidade ao campo magnético terrestre das abelhas

Abstract

In this work, a method bioinspired in the memory of the magnetic field of bees applied in a precision pollination rover is developed. The method calculates sharpening characteristics by entropy and Laplacian variance of color-segmented images in the HSV system in real time. A complementary positioning method was developed based on the extraction of area features between active markers, analyzing color, noise and vibration characteristics of the probe in time and frequency, through the lateral image of the probe. From the observed results, it can be observed that the unsupervised method does not require previous calibration of the target dimensions, histogram and distances involved in the positioning. The algorithm showed lower sensitivity in the extraction of sharpness characteristics in relation to the number of edges and greater sensitivity to the gradient, allowing unforeseen operating scenarios, even in small sharpness variations, and robust response to local, temporal and geophysical variation of magnetic declination, no need for luminosity after scanning, with two degrees of freedom of rotation. In the developed system, the scanning operation and the search for the minimum variation between the sharpness of the object and the sounding system are carried out, without variation regarding the magnetic declination, even in environments with low light after the acquisition of the frames. The system can be used in long-distance target research, crane positioning, precision agriculture, among others. From the initial results, it can be estimated that the angular alignment error is at most 1 degree.