Paper No. 991043 An ASAE Meeting Presentation ACCURACY OF DGPS FOR GROUND APPLICATION IN PARALLEL SWATHS by J. P. MOLIN Professor Dept. of Rural Engineering, ESALQ, Un. of São Paulo Piracicaba, SP, Brazil E. R. S. RUIZ Graduate Student Agricultural Machinery, ESALQ, Un. of São Paulo Piracicaba, SP, Brazil Written for Presentation at the 1999 ASAE/CSAE-SCGR Annual International Meeting Sponsored by ASAE/CSAE-SCGR Sheraton Center Toronto Hotel Toronto, Ontario, Canada July 18-21, 1999 Summary: Alternatives have to be found to justify the cost of DGPS for activities like yield mapping and site specific application. The use of GPS in agriculture is something new in Brazil. The same is not true for lime, an important input for highly acid soils, especially in the new frontiers in the central part of the country. All lime is applied by pull-type dry spreaders or trucks with no guidance. This work aims to explore the quality of DGPS as a guiding system for ground applications like lime, in vast agricultural areas. A methodology was proposed and tests were run for measuring the quality of a light bar guiding system and no guidance for ground applications. Keywords: DGPS, Guidance, Parallel Swathing The author(s) is solely responsible for the content of this technical presentation. The technical presentation does not necessarily reflect the official position of ASAE, and its printing and distribution does not constitute an endorsement of views which may be expressed. Technical presentations are not subject to the formal peer review process by ASAE editorial committees; therefor, they are not to be presented as refereed publications. Quotations from this work should state that it is from a presentation made by (name of author) at the (listed) ASAE meeting. EXAMPLE From Author s Last Name, Initials. Title of Presentation. Presented at the Date and Title of meeting, Paper No. X. ASAE, 2950 Niles Road, St. Joseph, MI 49085-9659 USA. For information about security permission to reprint or reproduce a technical presentation, please address inquiries to ASAE. ASAE, 2950 Niles Road, St. Joseph, MI 49085-9659 USA Voice: 616.429.0300 FAX: 616.429.3852 E-Mail: hq@asae.org
ACCURACY OF DGPS FOR GROUND APPLICATION IN PARALLEL SWATHS 1 MOLIN, J.P. 2 and RUIZ, E.R.S 3 ABSTRACT A methodology was developed to measure the tractor track error in parallel swath applications. This methodology was used in a field test for comparing two guiding conditions: with and without light bar as guidance. The light bar was governed by DGPS and was installed in a tractor. Though results showed better results from the light bar when compared with the system without any kind of guidance, the resulted error was too high for a parallel swathing application. Light bar setup and operators lack of experience may have affected the results. INTRODUCTION The Global Positioning System (GPS) is being used in several ways in agriculture. Precision agriculture got a powerful growth because of GPS availability and popularization. In fact, it is redundant to say that GPS is the most important tool for precision agriculture. Parallel uses have been proposed for GPS in agriculture and one of those is the guiding system using light bar activated by DGPS. This system was originally proposed and developed for aerial applications, guiding the airplane through parallel swaths without any ground reference. Today it is already being used for ground straight parallel swathing and companies are already announcing a new generation of hardware and software capable of guiding in parallel curved lines. BUIK (1998), in New Zealand, tested DGPS guiding compared to foam. He observed that DGPS gave an average error of 0.57 m against the foam that resulted in an error of 1.61 m. VETTER (1995) tested a DGPS light bar system for ground applications of straight parallel swaths. He used sophisticated video equipment. The results showed that in 50% of the time the error was less than 0.38 m and in 90% of the time that error was less than 0.89 m. The maximum displacement in 23 km of tests was 1.88 m. A similar test was conducted by VETTER (1996) for aerial application, using the same equipment and the same methodology. He found that in 50% of the time the error was less than 0.43 m and in 90% of the time that error was under 0.94 m. In 8 km of testing flight, the maximum error was under 1.3 m. In a banana plantation in Colombia, tests were conducted comparing the traditional system of guiding airplanes by flags in the ground versus a light bar activated 1 Part of this work has been supported by FAPESP (São Paulo State Research Foundation). 2 Dept. of Rural Engineering, ESALQ, University of São Paulo, Av. Padua Dias, 11, Piracicaba, SP, 13418-900, Brazil. Phone: (019) 429-4165, Fax: (019) 433-0934, e-mail: jpmolin@carpa.ciagri.usp.br 3 Graduate Student, Agricultural Machinery, ESALQ/USP 2
by DGPS. The light bar showed to be lot more efficient by both economic and environment perspective (ARDILA, 1996). MOLIN (1998) tested the traditional flagging guiding system versus the light bar using DGPS in a sugarcane area. For both systems exposed to the same conditions in the same area, the light bar resulted in a lot better performance. An error under 2.9 m for the flagging system against 0.5 m was observed at 50% of the time. At 90% of the time that error was under 4.4 m for the flags and less than 1.4 m for the light bar. The method used for measuring was not totally unbiased because the GPS equipment collected the path. An assumption of normally distributed error was necessary for that. Light bar guiding is being used for aerial application in Brazil for more than five years. Presently had been seen attempts of using it for ground applications in sprayer booms and spreaders. Considering that this system is being offered and used without clear information about its performance, a methodology was proposed and tested for measuring the displacement or error of a light bar guiding versus no guiding system at all. This is the normal condition when spreading lime and fertilizer with spinner box machines. In addition, farmers have to know if they will be able to give more use of the high investment of GPS and the annual rate for differential signal that they pay just for yield mapping and, maybe VRT. The quality of DGPS guidance with light bars is not well known but may provide an important alternative for lime application with spreaders that does not use any kind of guidance yet. MATERIAL AND METHODS Initially this work proposed to develop a methodology that could be used for testing the accuracy of guidance for swath applications in the field. A device was proposed to allow a tractor to drive through parallel swaths simulating field operations and marking exactly its route. This device is basically made of a flat vertical disk fixed to an articulated fork that allows the disk to compensate for small angles generated by changes on the tractor s direction (Figure 1). The fork was adapted to a tool bar attached to the three-point hitch of the tractor and positioned right in the center of it. The disk makes a visible mark in the ground while the operator simulates operations that need parallel swathing. The equipment used for guidance in the tests is a SatLoc SwathStar that has on it a 10 channels GPS, a CPU, a power supply box, an alphanumeric keyboard, a display and a light bar. The light bar has two horizontal rows of 28 leds each. It was set up having the first five leds in the center adjusted for the minimum resolution available (0.30 m) and than the intervals were progressively incremented to up to 15.00 m, by 3.00 m intervals, with the final setup for the leds distance intervals presented on table 1. Differential correction signal was supplied by satellite using an OmniStar receiver. Table 1. Light bar set up representing a particular deviation from the correct direction for each of the leds available, departing from the center to each side of the bar. Number of leds on 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Deviation (m) 0.3 0.6 0.9 1.2 1.5 2.0 2.5 3.0 4.0 5.0 6.0 8.0 10.0 12.0 15.0 3
Figure 1. Equipment designed and built to make visible marks in the ground. The tests were conducted in a regular field with smooth topography. Two speeds were used during the tests: 2.20 m/s (7.92 km/h) and 3.75 m/s (13.50 km/h). As the objective was to test the performance of GPS guidance, the tests were run in two conditions: with a light bar activated by DGPS and without any reference. As the operator will influence the actual track, based on his personal decisions as he travels along a line, five operators were used in the tests and considered as blocks. Four of the operators had years of experience on agricultural machinery and the other one was a graduate student with some experience. None of them had any experience on light bar use for orientation. They just practiced for some time before filling comfortable for the tests. For each plot 200 m long, it was established a base line made of stakes 50 m apart. On each treatment the tractor was driven close to the base line first and than three more passes were made on 5.0 m wide swaths. In the case of the light bar treatments, the equipment was set for this width and on those treatments without guidance, the operator was responsible for deciding and establishing the tractor s position, as it is made on operations like spreading. After passing through a sequence of one treatment, the heads of the plot were marked with stakes in every 5.0 m, fixing the actual starting position of each pass. A theodolite was used to establish the desired line on each pass. With the use of a measuring tape (Figure 2), the lines were measured in intervals of 5.0 m and in each of those points, in the perpendicular direction, it was measured the deviation between the actual and the desired pass. RESULTS AND DISCUSSION A good mark was obtained when using the disk in the ground covered with some residue and grass. The methodology proposed and used shown to be simple and useful, offering results at a centimeter level. 4
Figure 2. Measuring procedure between visible marks and desired line A set series of 40 readings for each pass with three replications was used for representing the deviation or error analysis using frequency and probability distribution. Table 2 presents the variance analysis of the data showing that the equipment was the most important source of variation. The average error for the four treatments tested and the mean comparison test are shown on table 3. Table 2. Variance analysis results Source of variance F test Blocks (operators) 6.08* Speed (2.20 m/s and 3.75 m/s) 21.17* Equipment (with and without DGPS) 292.30* Speed x Equipment 0.74 Numbers followed by * indicate F test significant at 1% Table 3. Mean errors observed for the two speeds tested with and without the light bar. Speed (m/s) Equipment Standard deviation Mean error (m) 2.20 without light bar 2.17 2.91 a 1 3.75 without light bar 2.03 2.57 a 2.20 with light bar 1.28 1.42 b 3.75 with light bar 0.71 0.91 c 1 Means followed by the same letter are not significantly different (Tukey 1%). 5
Figures 3 and 4 present the frequency distribution of the errors that were measured for the two speeds tested. The results represent the average of five operators showing a large deviation from the desired direction in both conditions, without using any reference and with the use of the light bar. Despite that, a significant difference was obtained between the two conditions, showing better results on the DGPS treatments. The same is for the speeds tested, showing significantly better results at the highest speed (3.75 m/s) and indicating that the operators tend to drive in a more smooth pattern at higher speeds because of having less time for reacting. Figure 3. Frequency distribution of error observed with and without the use of a light bar at traveling speed of 2.20 m/s. ERROR AT 2.20 m/s 25 Frequency (%) 20 15 10 5 0-10 -8-6 -4-2 0 2 4 6 8 10 ErroR (m) with DGPS w ithout DGPS Figure 4. Frequency distribution of error observed with and without the use of a light bar at traveling speed of 3.75 m/s. ERROR AT 3.75 m/s 30 Frequency (%) 25 20 15 10 5 0-10 -8-6 -4-2 0 2 4 6 8 10 Error (m) with DGPS w ithout DGPS 6
On figures 5 and 6 it is possible to see the probabilities associated with the errors. They show the same clear difference between the two systems: no reference and light bar. Considering the speed levels and the probability errors observed when using light bar, if compared to the result obtained by VETTER (1995) when using a similar specification on the equipment, it indicates the effect of operator s reaction related to the available time. At 2.20 m/s, the deviation was lower than 1.10 m 50% of the time and lower than 3.10 m 90% of the time. When considering the results from 3.75 m/s, 50% of the time the error was under 0.80 m and 90% of the time that error was under 2.10 m. Figure 5. Error probability observed with and without the use of a light bar at traveling speed of 2.20 m/s. ERROR PROBABILITY AT 2.20 m/s Probability (%) 100 80 60 40 20 0 0 2 4 6 8 10 Error (m) with DGPS without DGPS Figure 6. Error probability observed with and without the use of a light bar at traveling speed of 3.75 m/s. ERROR PROBABILITY AT 3.75 m/s Probability (%) 100 80 60 40 20 0 0 2 4 6 8 10 Error (m) with DGPS without DGPS 7
Using light bar for guiding showed to be lot better than using nothing as normally farmers proceed on operations like lime and fertilizer spreading. Nevertheless the results are not precise enough if considered the swath width of lime and fertilizer when using spinner box machines. The light bar setup has to be considered in any subsequent test as the 0.30 m interval for the first leds was to narrow for the DGPS accuracy. It forced the operators to change the direction so frequently that it may affected negatively the results. CONCLUSIONS The methodology proposed showed to be very efficient and simple, offering results at a centimeter level. Using light bar as guiding system resulted in less error than operating the tractor without any kind of guidance. The error was affected by the traveling speed, decreasing between 2.20 m/s and 3.75 m/s. The light bar setup and lack of practice from the operators may have influenced the results. Parallel swathing without any guiding system resulted in high errors not considered by farmers. ACKNOWLEDGEMENT This work was made possible with the use of SatLoc and Omnistar equipment and services to whom we thank. REFERENCES ARDILA, M. J. 1996. An appealing alternative: aerial guidance over Colombia s banana field. Precision Farming. December 1996. p. 10-15. BUICK, R. How precise are parallel swathing systems? Modern Agriculture. 1998. P.32-34. MOLIN, J. P. 1998. Análise comparativa de sistemas de orientação com DGPS e com bandeiras para aeronaves agrícolas. XXVII Brazilian Congress of Agricultural Engineering. Proceedings. Poços de Caldas. SBEA/UFLA. V.3, p. 64-66. VETTER, A.A. Quantitative evaluation of DGPS guidance for ground-based agricultural applications. Applied Engineering in Agriculture. 1995. American Society of Agricultural Engineers. V: 11(3): 459-464. VETTER, A.A. Quantitative evaluation of DGPS guidance for arial agricultural applications. Applied Engineering in Agriculture. 1996. American Society of Agricultural Engineers. V: 12(5): 611-616. 8