Vision system for coconut farm cable robot

In many countries, robots and automation techniques are being introduced in agriculture farms to reduce the human labour and to improve the yield. However, such technological initiatives are still lacking in India, although it is the leading producer of many vegetables and fruits, for example, coconuts. Some of the activities carried out in a coconut farm that requires human labor are coconut dehusking, loading and unloading of coconuts. Automating these activities in a coconut farm would require a robotic system to pick and transport coconuts, for which the primary need would be to detect coconuts in those environments under natural lighting conditions. Towards this, the work in this paper tests for the applicability of three most used computer vision based object detection approaches namely, Local Binary Pattern (LBP) cascade, Histogram of Oriented Gradients (HOG) cascade and Haar — like cascade in coconut detection. This vision system would enable any field robot to automate the tasks in coconut farms without human assistance. A comparative analysis using confusion matrix is carried on these three approaches. It is observed that Haar-like features provided comparably better results among all the three features, in terms of hit rate and precision.