Development of Automated Guidance Tracking Sensor System Based on Laser Distance Sensors
Joon-Yong Kim1, Hak-Jin Kim1,2, Sung-Bo Shim3, Soo-Hyun Park4, Jung-Hun Kim5, Young-Joo Kim6*
1Dept. of Biosystems Engineering, Seoul National University, Seoul, 08826, Korea
2Research Institute for Agriculture and Life Sciences, Seoul National University, Seoul, 08826, Korea
3Upland-Field Machinery Development Research Center, Kyungpook National University, 80 Daehakro, Buk-gu, Daegu 41566, Korea
4Natural Products Research Center, Convergence Research Center for Smart Farm solution, KIST, Ganneung-si, Gangwon-do, 25451, Korea
5R&D Institute, Tong Yang Moolsan Co., LTD, Gongju-si, Chungcheongnam-do, 32530, Korea
6Convergence Agricultural Machinery R&D Group, Korea Institute of Industrial Technology, Gimje-si, Jeollabuk-do, 54325, Korea
Received: October 4th, 2016; Revised: November 18th, 2016; Accepted: November 25th, 2016
Purpose: Automated guidance systems (AGSs) for mobile farm machinery have several advantages over manual operation in the crop production industry. Many researchers and companies have tried to develop such a system. However, it is not easy to evaluate the performance of an AGS because there is no established device used to evaluate it that complies with the ISO 12188 standard. The objective of this study was to develop a tracking sensor system using five laser distance measurement sensors. Methods: One sensor-for long-range distance measurement-was used to measure travel distance and velocity. The other four sensors-for mid-range distance measurement-were used to measure lateral deviation. Stationary, manual driving, and A-B line tests were conducted, and the results were compared with the real-time kinematic differential global positioning system (RTK-DGPS) signal used by the AGS. Results: For the stationary test, the average error of the tracking sensor system was 1.99 mm, and the average error of the RTK-DGPS was 15.19 mm. For the two types of driving tests, the data trends were similar. A comparison of the changes in lateral deviation showed that the data stability of the developed tracking system was better. Conclusions: Although the tracking system was not capable of measuring long travel distances under strong sunlight illumination because of the long-range sensor’s limitations, this dilemma could be overcome using a higher-performance sensor.
Keywords: Automated guidance system, ISO 12188-2, Laser distance sensor, Track error, Tracking sensor
https://doi.org/10.5307/JBE.2016.41.4.319 eISSN : 2234-1862
pISSN : 1738-1266
*Corresponding author: Young-Joo Kim Tel: +82-63-920-1277; Fax: +82-63-920-1280 E-mail: ojoo@kitech.re.kr
Introduction
Automated guidance systems (AGSs) have several advantages in crop production, one of which is that they can reduce operator fatigue. For example, it is possible to operate farm machinery using an AGS in conditions of poor visibility where a human operator would be incapable of operating the
machinery. In addition, they can reduce gaps and overlaps during harvesting and chemical application.
Various types of AGSs have been developed and distributed in the modern crop industry. Agricultural vehicle guidance technologies include laser-based sensors, inertial sensors, dead reckoning, machine vision, and global navigation satellite systems (GNSSs) (Li et al., 2009). Furthermore, various technologies such as computer simulations have been used to improve AGSs (Han et al., 2015).
The accuracy and repeatability of an AGS are important.
When it turns off the row of the field, the crop would be damaged, because it is used in harvest operation or chemical application. Researchers have conducted studies to evaluate the performance of AGSs. Ehsani et al. (2002) introduced a method for evaluating the accuracy of guidance systems and evaluated six commercially available guidance systems. In this study, the accuracy was measured by calculating the mean of the errors, where the error was defined as the actual in-field position of the vehicle from the desired position while driving in a straight line. Ehrl et al. (2004) investigated the accuracy of an AGS using two independent geodetic real-time kinematic differential global positioning system (RTK-DGPS) rovers. Their aim was to evaluate the AGS’s accuracy under typical agricultural conditions. Baio (2012) evaluated the accuracy, cane loss, and operational field efficiency achieved by an AGS used to guide a sugarcane harvester over the field compared to a manually guided machine.
Various evaluation methods have also been discussed.
Davood et al. (2006) proposed a method for evaluating a guidance system that involved using an RTK-DGPS to record the location of a tractor, and analyzing the frequencies of recorded data. Rovira-Mas et al. (2008) suggested a method to evaluate the behavior of any automatically driven agricultural vehicle traveling along paths of any curvature. Randy Ross (2011) discussed a method to create a single graph that easily relates the field efficiency gains expected for a particular guidance system used in farming operations.
Additionally, a number of evaluation devices have been developed. Han et al. (2004) developed a method to evaluate the dynamic position accuracy of a GPS using a linear parallel-tracking application. It was proposed that the pass-to-pass average error was the most important metric. Harbuck et al. (2006) assessed different AGSs using a non-GPS-based device over various time periods. A tracking prism was installed on the towing hitch at the rear of the tractor, and the cross-track error (XTE) and pass-to-pass error were measured. Adamchuk et al. (2007) used a linear potentiometer for a GPS-based auto-guidance test program.
The sensor was installed in a cart that the target tractor pulled. It could measure the horizontal position of the cart with 20-mm resolution as it repeatedly passed over a series of stationary metal triggers installed on the surface of the test track. Easterly et al. (2010) developed a performance testing system for auto-guidance using a vision sensor system that was mounted on the rear of the tested tractor.
It tracked a permanent reference line on the test track.
As previously discussed, various researchers have used different test procedures and different devices for AGS evaluation. Their differences led to the necessity of additional research to compare the performance of different approaches.
Adamchuk et al. (2007) and Easterly et al. (2010) pointed out the need for a standardized procedure to test and report the performance of AGSs. American Society of Agricultural and Biological Engineers (ASABE) suggested ASABE X605, which is a draft for satellite-based AGS testing during straight and level travel (ASABE, 2008). ISO 12188-2 is a standard for satellite-based auto-guidance testing based on ASABE X605 (ISO, 2012), which defines a tracking sensor. According to ASABE X605, a tracking sensor is an instrument or instrumented system designed to conduct repeated horizontal distance measurements required for XTE calculations, and it should not rely on a GNSS-based positioning device to perform such measurements (ASABE, 2008). The definition in ISO 12188-2 is slightly different. This standard defines a tracking sensor as an instrument or instrument system designed to produce horizontal distance measurements required for error calculations that are at least ten times more accurate than the accuracy of the AGS being tested (ISO, 2012).
This study was initiated to develop an AGS for a tractor. To evaluate the AGS, a tracking sensor was needed. The objective of this study was to develop a tracking sensor system using laser distance measurement sensors (LDMSs) and to then evaluate the system. The tracking sensor system consisted of five LDMSs. One sensor-for long-range distance measurements- was used to measure travel distance and velocity. The other four sensors-for mid-range distance measurements-were used for lateral deviation measurements.
Materials and Methods AGS Standard
The standard referred to in this study states that the representative vehicle point (RVP) is a fixed point on a vehicle used to represent the vehicle’s location at any time.
The tracking sensor should be used to make relative distance measurements with respect to the RVP. The sensor can measure the distance between the permanent surface marker and the RVP. The track error (TE) is an error estimate determined as the horizontal distance between the A-B line, which is the desired path to follow, and any
Figure 1. Definition of TE and XTE.
Figure 2. Installation positions of LDMSs. Figure 3. LDMS for lateral error measurement.
given position of the actual RVP travel path. The XTE is the difference of the TE on the prior RVP travel path and the TE on the later RVP travel path. The concept of the TE and XTE is shown in Figure 1.
Concept of tracking sensor system
Although the RVP can be any point on the target tractor (or, optionally, on an apparatus attached to the tractor), the RVP in this study was located in the center of four LDMS sensors that were installed at the front and at the back ends of the tractor. The proposed tracking sensor system used five LDMSs to measure directional distance and lateral distance. An LDMS for long-distance measurement was installed behind the tractor and measured the traveling distance and velocity, as shown in Figure 2.
Figure 2 also shows the installation position of the four LDMSs. Two LDMSs were installed at the front, and the others were installed at the rear. The sensors measured the relative distances between the permanent surface marker and the sensors. An LDMS at the front and an LDMS at the rear measured the distance during initial travel, and the others measured the distance during subsequent travel. The distances were used to calculate the heading direction (yaw) and the lateral distance of the tractor. The standards referred to in the study focus on AGS performance while traveling straight over a level surface. The roll and
pitch of the tractor were not considered.
When L1 and L2, which are the distances between the permanent surface marker and the sensors, were known, the heading angle of the tractor to the marker could be calculated using Eq. (1).
tan
(1)where θ is the heading angle of the tractor to the reference wall, L1 is the distance between the right-side LDMS at the front and the permanent surface marker, L2 is the distance between the right-side LDMS at the rear and the permanent surface marker, and W is the distance between the LDMSs at the front and the rear.
Let L be the distance between the permanent surface marker and the A-B line and let D be the lateral offset between the LDMSs and the RVP. The lateral error could then be calculated using Eq. (2).
cos
(2)where L is the distance between the permanent surface marker and the A-B line, D is the lateral offset between the LDMSs and the RVP, and TE is the track error.
Tracking sensor system
As shown in Figures 2a and 2b, the five LDMSs were installed at the front and rear of the target tractor. Four mid-range distance sensors (DT50, SICK AG, Germany) and a long-range distance sensor (LT2000-SO, MODULOC
Table 1. The specifications of the sensors used in the study
Items Values
Mid-range distance sensors
(DT-50)
Measuring range 200 mm-10,000 mm, 90% remission
Resolution 1 mm
Repeatability 5 mm/2.5 mm (fast/slow)
Accuracy ±10 mm4)
Long-range distance sensor
(LT2000-SO)
Measuring range 0.2 m-30 m
Resolution 0.1 mm
Repeatability ±0.5 mm
Accuracy ±3 mm for 15-30°C
Figure 4. View of tracking sensor system.
Table 2. The specifications of the data acquisition device used in the study
Items Values
Signal range ±10 V
Sample rate 500 kS/s/ch
Resolution 16-Bit
Channels 4 differential
Isolation 60 VDC Ch-Ch
Table 3. The specifications of the tractor used in the study
Items Values
Engine
Type 4 cycle, 4 cylinder, Diesel
Power/rotational speed (kW/rpm) 58.8/2,400
Displacement (cc) 3,300
Dimension Length × Width × Height (mm) 3,940 × 1,940 × 2,675
Wheel base (mm) 2,200
T/M Front 24
Rear 24
Speed Forward (km/h) 0.25-35.6
Backward (km/h) 0.24-34.7
PTO speed (rpm) 540/750/1,000
Table 4. The specifications of the GPS used in the study
Items Values
Base GPS (FlexPak-G2-V2)
Horizontal position accuracy (RMS)
Single point L1 1.5 m
Single point L1/L2 1.2 m
SBAS 0.6 m
DGPS 0.4 m
RT-20 0.2 m
RT-2 1 cm+1 ppm
Maximum data rate 50 Hz
Channel count 72
Rover GPS with an IMU (SPAN-CPT)
Horizontal position accuracy (RMS)
Single point L1/L2 1.2 m
SBAS 0.6 m
DGPS 0.4 m
PPP 0.04 m
RTK 1 cm+1 ppm
Data rate
GPS measurement 20 Hz
GPS position 20 Hz
IMU measurement 100 Hz
INS solution Up to 100 Hz
Figure 5. Installation of reference wall as permanent surface marker Installation of reference wall as permanent surface marker.
Control Systems Ltd., USA) were used (Table 1). The mid-range distance sensor was an analog output type and could measure distances ranging from 20 cm to 10 m. Its response time and repeatability were 20-30 ms and 2.5-5 mm, respectively. The long-range distance sensor supported RS232 or RS422/RS485 communication and provided an accuracy of ±2 to ±5 mm. It could measure the distance to a white target 100 m away.
Two brackets were constructed to attach the sensors.
Two LDMSs for mid-range distance were attached to the front guide. The other two LDMSs for mid-range distance and an LDMS for long-range distance were attached to the rear guide. Figure 4 shows the front and rear ends after
installation of the LDMSs. The leveler shows that the sensors were installed horizontally.
As shown in Table 2, the mid-range distance sensors were connected with a data acquisition device (NI 9222, National Instruments Corporation, USA). The long-range distance sensor was directly connected to the main computer of the tracking sensor system. The software was implemented using LabView (Ver. 2011, National Instruments Corporation, USA).
A first-order low-pass filter was employed to reduce noise.
The measured data and processed results were saved in a new file each time the sftware ran.
In order to build a reference wall as permanent surface marker, a recycled plastic panel (RPP) was used. The RPP, which can be used to reduce noise from construction sites, comprises a flat and long wall. As shown in Figure 5, the RPP wall was installed without any bumps. The reference wall was installed at the high-tech agricultural machinery support center of the Korea Institute of Industrial Technology in Gimje, Jeollabukdo, Korea. Figure 5 shows the aerial view of the center and the position of the reference wall. The length and height of the reference wall were 50 m and 1 m, respectively.
Tested system description
The tested system was initially developed on a tractor (TX 803, TongYangMoolSan Co., Ltd., Korea). It used a base GPS (FlexPak-G2-V2, NovAtel Inc., Canada) and a rover GPS with an inertial measurement unit (IMU) (SPAN-CPT,
Figure 6. Results of stationary test.
Figure 7. Results of manual driving test.
NovAtel, Inc., Canada) for navigation. The IMU was installed 20 cm to the right of the center line. The system was used to control the steering, the clutch, and the brakes and uses a controller area network (CAN). An embedded box computer (MXC-6201D/M4G, ADLINK Technology, Inc., Taiwan) and a touch-screen monitor were also installed. The controlling software was implemented using LabView (Ver. 2011, National Instruments Corporation, USA). Han et al. (2013) reported the path generation and tracking algorithm used.
Experimental design
To evaluate the tracking sensor system, three different tests were designed. The first test was a stationary test to compare the position data measured by the tracking sensor system with the position data measured by the GPS and IMU when the target tractor was stationary after starting up.
Each measurement was executed for 1 min, and any initial data that was unstable during initialization was removed.
This test was repeated three times at different places on three different days.
The second and third tests were driving tests. The second test was manual driving without steering. To reduce disturbances from other conditions, the tractor was manually driven with the steering wheel being held constant to prevent the tractor from turning (the tractor went straight). The third test was executed using the developed AGS.
The standards recommend using three different speeds:
slow (0.1±0.05 m/s or the minimum recommended speed for the vehicle in use), medium (2.5±0.2 m/s), and fast (5±0.2 m/s) (ASABE, 2008; ISO, 2012). The tests were executed using the slow speed because the tested tractor could operate only at a slow speed when it was tested. The tested track was an asphalt pavement surface with a slope of less than 3%. The length of the test course was less than 50 m because it depended on the length of the reference wall. Each test was repeated three times.
Results and Discussion Stationary test results
The RTK-DGPS provided absolute position data, but the tracking sensor system provided relative position data. To compare these, a coordinate transformation was needed.
There were a total of nine tests. The average distance of the RTK-DGPS was 15.19 mm, and its standard deviation was 6.87 mm. The average distance of the tracking sensor
system was 1.99 mm, and its standard deviation was 0.93 mm. The tracking sensor system was more precise than the GPS when the tractor was stationary.
Figure 6 shows the results of the stationary test. Each spot represents a relative position from the origin, which was set using the average value of the GPS signal and the tracking sensor system. The GPS spots were spread over the field, but the spots of the tracking sensor system were gathered near the origin. One important characteristic of the GPS signal is that it made two separate groups. The left-side island data were collected during the initial 18.2 s.
This shows that the GPS signal bounced.
Manual driving test results
For the manual driving test, the steering wheel was fixed, and the tractor was driven with a fixed velocity.
Although the steering wheel was fixed, the trajectories were not straight. There are several factors that can affect
Figure 8. Results of A-B line test.
Figure 9. Lateral changes of A-B line test.
Figure 10. Measurement errors of long-range distance sensor.
the trajectories, such as the shapes of the tires, the surface of the test course, the vibration of the tractor, and mistakes made by the driver. Figure 7 shows the results of the manual driving test. Figure 7(a) shows the trajectories measured by the GPS and the tracking sensor system.
Their trends were quite similar, but there were still small differences.
The GPS signal had several signal bounces. Figure 7(b) shows the lateral differences between the neighboring two points and emphasizes the signal bounce. At 5, 16, and 21 m, the differences of the GPS signal were bigger than those of the tracking sensor system signal.
A-B line test results
The tracking sensor system was tested using the AGS for the A-B line test. Two points (A, B) were given as a path.
The two points were selected using the average value of the DGPS signals. The distance between A and B was 44.72 m. Figure 8 shows the results of the A-B line test. There are four plots, and each plot shows movement corresponding to 10 m. During the first 10 m, both signals were smooth.
However, there were several jumps in the GPS signal of the other plots.
This trend was more pronounced in Figure 9, which shows the changes of the lateral positions. The DGPS signal showed several sudden changes, but the signal of the tracking system was stable. The change range of the DGPS signal was from -0.0265 to 0.0194 m. However, the change range of the tracking sensor system was from -0.0054 to 0.0046 m. The range of the tracking sensor system was much smaller, which means that the signal of the tracking sensor system was more stable compared with the DGPS signal.
Discussion
This system has several advantages and disadvantages.
This system’s concept was similar to those of the two standards. This system was capable of measuring the relative position of the AGS from the reference wall. In addition, the tracking sensor system could collect data at 20 Hz-faster than the 10 Hz recommended in both standards.
The measurable lateral range of this system was wide, although it depended on the heading angle. In addition, it could measure the directional distance. This means that the velocity of the tractor could be calculated, and this system could be applied in a variable-speed test.
On a sunny day, the long-range distance sensor was disturbed by sunlight. Although the specifications of the
sensor stated that the range was 100 m, the sensor occasionally provided errors when the tractor moved over 35 m. Figure 10 shows one example of the error that occurred over 35 m. This phenomenon did not occur at night. Although the sensor could be used under weak sunlight, a higher-level sensor or a radar sensor would be preferable to measure the velocity of the tractor. Further research could be used to evaluate the effectiveness of a radar sensor to measure the velocity of a vehicle instead of a long-range distance sensor.
Conclusions
The evaluation of AGSs has been an important issue since they were first developed. Nowadays, a standard -ISO 12188-2- is available. In this standard, the XTE is the main parameter, but it is not easy to measure because there is no standard device for this. In this study, a tracking sensor system, which could measure TE, was developed using five LDMSs. One sensor-for long-range distance measurement-was used to measure travel distance and velocity. The other four sensors-for mid-range distance measurement-were used for lateral deviation measurement.
Three tests were conducted, and the results were compared with the RTK-DGPS signal. For the stationary test, the average error of the tracking sensor system was 1.99 mm, and the average error of the RTK-DGPS was 15.19 mm.
The trajectory measured by the tracking sensor system was more stable for both the manual driving test and the A-B line test. Although there were some disadvantages resulting from the sensor specifications, the tracking sensor system could effectively be used to evaluate the AGS based on the aforementioned standard.
Conflict of Interest
The authors have no conflicting financial or other interests.
Acknowledgments
This work was supported by the Technology Innovation Program (Project 10039936 for the development of an intelligent steering control system for an auto-guidance agricultural tractor) funded by the Ministry of Knowledge Economy (MKE, Korea).
Nomenclature
TE : Track error XTE : Cross-track error
References
Adamchuk, V., R. Hoy, G. Meyer, M. Kocher and J. Stafford 2007.
GPS-based auto-guidance test program development, the Sixth European Conference on Precision Agriculture.
Wageningen Academic Publishers, Wageningen, The Netherlands 425-432.
ASABE, 2008. X605 Satellite-Based Auto-Guidance Systems Testing During Straight and Level Travel.
Baio, F. H. R. 2012. Evaluation of an auto-guidance system operating on a sugar cane harvester. Precision Agric 13:141- 147.
Davood, K., D. M. Danny and E. Reza 2006. A New Methodology for Evaluating Guidance Systems for Agricultural Vehicles.
ASAE Paper No. 06148.
Easterly, D. R., V. I. Adamchuk, M. F. Kocher and R. M. Hoy 2010.
Using a vision sensor system for performance testing of satellite-based tractor auto-guidance. Computers and Electronics in Agriculture 72:107-118.
Ehrl, M., W. V. Stempfhuber, M. R. Demmel, M. Kainz, H.
Auernhammer 2004. AutoTrac-Accuracy of a RTK DGPS based Autonomous Vehicle Guidance System under Field Conditions, Automation Technology for Off-Road Equipment.
ASABE, Kyoto, Japan, 274-282.
Ehsani, M. R., S. Matthew, Dr. T. W. Joel and T. L. Zimmerman 2002. A Method of Evaluating Different Guidance Systems.
ASAE Paper No. 021155. St. Joseph, MI.: ASAE.
Han, S., Q. Zhang, H. Noh and B. Shin 2004. A dynamic performance evaluation method for DGPS receivers under linear parallel-tracking applications. TRANSACTIONS- AMERICAN SOCIETY OF AGRICULTURAL ENGINEERS 47:321-330.
Han, X.-Z., H. -J. Kim, H. -C. Moon, H. -J. Woo, J. -H. Kim and Y.-J. Kim 2013. Development of a Path Generation and Tracking Algorithm for a Korean Auto-guidance Tillage Tractor. Journal of Biosystems Engineering 38:1-8.
Han, X. Z., H. J. Kim, J. Y. Kim, S. Y. Yi, H. C. Moon, J. H. Kim and Y.
J. Kim 2015. Path-tracking simulation and field tests for an auto-guidance tillage tractor for a paddy field. Computers and Electronics in Agriculture 112:161-171.
Harbuck, T. L., J. P. Fulton, T. P. McDonald and C. J. Brodbeck 2006. Evaluation of GPS autoguidance systems over varying time periods. ASAE Paper No. 061042. St. Joseph, MI.: ASAE.
ISO, 2012. ISO 12188-2:2012 Tractors and machinery for agriculture and forestry-Test procedures for positioning and guidance systems in agriculture-Part 2: Testing of satellite-based auto-guidance systems during straight and level travel.
Li, M., K. Imou, K. Wakabayashi and S. Yokoyama, 2009. Review of research on agricultural vehicle autonomous guidance.
International Journal of Agricultural and Biological Engineering 2:1-16.
Randy Ross, P., 2011. A General Method to Illustrate the Different Field Efficiency Gains of Guidance Systems. ASABE Paper No.
1110848. St. Joseph, MI.: ASABE.
Rovira-Mas, F., H. Shufeng, and J. F. Reid 2008. Evaluation of automatically steered agricultural vehicles, Position, Location and Navigation Symposium, 2008 IEEE/ION, 473-478.
