Agricultural tools guidance assistance by using machine vision

This paper presents a regulation mechanism aiming to position agricultural tools relatively to the previous lines, while sowing or harvesting. The sowing rows were revealed by a background correction, the background being obtained thanks to a median rank filter. The method was found efficient in eliminating the shadows. For the crop rows (chicory rows), a neural network was used to localise the plants. While the petiole and the leaves were easily separated from the soil, the chicory root and the soil having about the same colour and the lighting condition varying widely, it was more difficult to obtain a good contrast between those parts, which leaves place for some improvements. The adapted Hough transform consisted in computing one transform for each line in the cluster with, for reference, the position and direction of the theoretical position of the row. The different transforms were then added. The position was used in a feedback regulation loop. An articulated mechanism was used to ensure the lateral displacement of the tool relatively to the tractor. The behaviour of the whole outfit was studied during several field tests. The standard deviation of the error, measured as the difference between the observed inter-row distance and its theoretica value, was of 23 mm for sowing and 31 mm for harvesting and its amplitude was less than 100 mm for sowing and less than 115 mm during the harvest, which was sufficient to fulfil the requirements of the application. Sources of systematic errors were also identified as linked to the geometric considerations. Their correction requires an accurate mounting of the camera, which may be possible for a serial montage.