Automatic Guidance of a Tractor Using Computer Vision
This paper presents a computer vision guidance system for agricultural vehicles. This system is based on a segmentation algorithm that uses an optimum threshold function in terms of minimum quadratic value over a discriminant based on the Fisher lineal di
- PDF / 13,051,072 Bytes
- 8 Pages / 430 x 660 pts Page_size
- 52 Downloads / 180 Views
ract. This paper presents a computer vision guidance system for agricultural vehicles. This system is based on a segmentation algorithm that uses an optimum threshold function in terms of minimum quadratic value over a discriminant based on the Fisher lineal discriminant. This system has achieved not only very interesting results in the sense of segmentation, but it has also guided successfully a vehicle in a real world environment.
1 Introduction This paper describes an automatic agricultural computer vision guided system. Precision Agriculture is defined as the application of technical advances to agriculture; this field has grown rapidly in recent years because it offers a cheaper and more effective way of working the land. When using precision techniques it is more appropriate to use a local rather than a global reference due to benefits as position independence, a priori terrain knowledge independence and tolerance independence [1]. We have chosen a concrete local reference technique that has been rapidly developed for agricultural purpose: Computer Vision. The first tentative trials were direct applications of the traditional computer vision techniques like FFT convolution filters, basic border detection or region division techniques [1], [2]. The Hayashi and Fujii system based on border detection and Hough transform [3] and the system based on texture and intensity from the group of Chateau [4] could be some examples. Lately, some investigators like Reid and Searcy [5], Gerrish [6] and Klassen [7] have developed quite effective systems which were never actually tested guiding a vehicle. In recent years, some systems have achieved great results when guiding a vehicle in a very specific task, like the Ollis and Stentz algorithm [8] for harvesting or the Bulanon, Kataoka, Ota and Hiroma [9] guidance system to collect apples. However, there is still no effective solution to guide an agricultural vehicle with independence of the task.
2 Objectives As mentioned above, the main goal for this project is to develop the most generalpurpose solution possible. Thus its success must be measured in terms of the number A. Ghosh, R.K. De, and S.K. Pal (Eds.): PReMI 2007, LNCS 4815, pp. 169 – 176, 2007. © Springer-Verlag Berlin Heidelberg 2007
170
P. Moreno Matías and J. Gómez Gil
of agricultural tasks that can be guided by the system. The concrete effectiveness for each concrete task is less important. The decision of whether a task can be reasonably done or not with the system will depend on the relative distance between the actual division line between areas and the approximation of the algorithm for the segmentation algorithm, and the real distance between rows in the guidance algorithm.
3 Development As described, the system is based on a software application that decides the direction to be taken by the tractor and a hardware interface that communicates with the hardware guidance system. The most complex and interesting part of the system is the software application that implements the computer vision logic. The s
Data Loading...
