Submission Template for Student Design Competition 2014
University: Seoul National University, South Korea
Team Members (with year of graduation): Xiong-Zhe Han (2nd year of Ph.D. course)
Faculty Advisers: Prof. Hak-Jin Kim and Dr. Hee-Chang Moon
Email Address: firstname.lastname@example.org
Submission Language: English
Title: Path Tracking Simulation and Field Tests for an Auto-Guidance Tillage Tractor System
Description: In this project, we developed a computer simulator capable of virtually testing the motion of a tractor based on a waypoint-based path tracking algorithm and tractor dynamics in a 3D graphic environment. A path tracking algorithm for the tractor was developed with the LabVIEW 2013. The tractor motion was simulated based on the dynamic model of a vehicle that considered the effects of tire slippage and side force imparted by soil. The validity of the computer simulation was confirmed using an auto-guided tractor equipped with an RTK-GPS system, the path tracking algorithm, and an integrated driving controller implemented with a NI Compact-RIO. In an arable field (90 m x 25 m), the auto guided tractor followed the predefined path including C-shape headland turning with acceptable tracking, showing RMS lateral errors ranging from 3.8 to 12.8 cm on the straight paths. However, the RMS lateral errors obtained on the curved paths increased to 100 cm when the tractor traveled in a wet sub-field with a moisture content > 30 %, requiring more accurate estimation of sliding parameters.
NI Compact Rio (sbRIO-9620)
FlexPak-G2-V2 (GPS-based receiver)
SPAN-CPT (GPS-rover receiver)
701-GLL (GPS antenna)
ADL Vantage Pro ( Modem)
Visual studio 2008
When considering the use of an auto-guidance tractor in Korean paddy fields especially for rice cultivation, tire slippage and headland turning method are two main factors that may affect the accuracy of path tracking because the soils of paddy fields are generally wet, and the field size is small ranging from 0.3 to 1.0 ha as compared to USA corn and soybean fields. Therefore, the successful adaptation of an auto-guidance tractor in Korean paddy fields depends on the ability to accurately guiding the tractor along the desired path using a robust path tracking algorithm. Such a path tracking algorithm needs to be developed based on accurate estimation of vehicle motion on curved paths in the presence of sliding especially when the tractor moves from the current working row to the next one such as headland turning.
To simulate the motion of an autonomous tractor being operated based on a path tracking algorithm, a simulation program was designed, as shown in Fig. 1. The simulator architecture consisted of a client and a server. In the client environment, a desired path consisting of waypoints was set as an input, and the tractor steering angles needed to follow the desired path were calculated using the developed path tracking algorithm by comparing target points, i.e., waypoints, with actual positions obtained from the server environment. The path tracking algorithm was developed with the LabVIEW program. In the server environment, the tractor simulation was performed using a motion engine developed with the RoboticsLab program (Simlab Co, Seoul, Korea) based on the tractor dynamics, and the motion status of the tractor, i.e., the vehicle's position and attitude, was obtained in the simulation environment using a GPS and gyro sensors installed on the roof of the tractor. The information was sent back to the client. A TCP/IP communication mode was used for transmitting the package data between the client and server computers.
Fig. 1 Simulator architecture consisting of server and client environments
Development of the Tractor Driving Simulator
Figure 2 shows the driving simulator developed in the study in order to virtually test the motion of an auto-guidance tractor based on commands provided by the proposed navigation algorithm. Figure 2(left) indicates the user interface (UI) that was developed with the LabVIEW program to determine information about the vehicle path, predefined in RDDF format to include waypoints, LBO, desired velocity, and the status of the PTO operation. GPS signals received from the RTK-GPS system and the steering angles, yaw rates, and slip angles were calculated using the path tracking algorithm, and the kinematic vehicle model can be displayed in real-time. In addition, a map showing the tractor path can be automatically created and displayed along with the predefined waypoints. Figure 2(right) shows the motion of the tractor simulated in a 3D graphic environment.
a ) Tracking algorithm module (client) b) Simulation engine (server)
Fig. 2 Simulator environments
Construction of an Auto-Guidance Tractor System
The test platform used for the auto-guidance testing was a Tongyang Model TX803 tractor (Tongyang Moolsan Co., Seoul, South Korea). As shown in Fig. 3, the tractor was modified as an auto-guidance tractor system equipped with an RTK-GPS system for position determination, a navigation computer for path tracking, an integrated driving controller for steering and traveling, and electric actuators for steering and gear shifting. The RTK-GPS system was equipped with a GNSS receiver (FlexPak-G2-V2, Novatel Inc., Canada) as a ground base station and a GNSS-INS receiver (SPAN-CPT, Novatel Inc., Canada) as a rover in order to locate the absolute position of the tractor with a positioning accuracy of up to +2 cm. These data were transmitted from the ground base station at 9600 bits/sec through two RF modems (ADL Vantage Pro, Pacific Crest, CA, USA); the base station was located approximately 100 m from the tractor. The heading angle of the tractor was measured using an IMU unit installed in the SPAN-CPT enclosure receiver. The absolute position and heading angle data were transferred to the navigation computer (MXC-6201D, M2G, Taiwan) running the Windows operating system through an RS232C interface at a rate of 20 Hz. The steering angles of the tires at each location were generated using the path tracking algorithm developed to follow a predefined tillage map with C-shape headland turning. The traveling velocity of the tractor was automatically read from the path planning map in RDDF. The steering angles of the front tires and travel velocity of the tractor were sent to the integrated driving controller that was developed with a single board (sbRIO-9606, National Instruments, USA) via CAN. The controller operated a steering motor installed on the steering column using a torque magnitude signal based on the error between the desired steering angle and the measured heading angle.
Fig. 3 Hardware architecture of the auto-guidance tractor system
Results of Simulator Validation
Figure 4 shows the results of the validation test on the computer simulation. As shown in Fig. 4(right), the tractor trajectories obtained from computer simulation and the paved road tests were similar, indicating that the computer simulator showed acceptable performance regardless of the path shape.
Fig. 4 Tractor platform and tracking trajectories obtained in the study
Results of Field Tests using the Auto-guided Tractor
Figure 5 shows the auto-guided tractor equipped with a rotary tiller being tested in a test field. As shown in the figure, the field was flat with some areas of high moisture content. The rotary tiller was operated when the tractor traveled along straight paths.
Fig. 5 View of the auto-guidance tractor system (left) tested in a paddy field (right) located in Korea.
Figure 6 shows the trajectories of the autonomous tractor based on the developed tracking algorithm in an arable field. The tractor traveled in a sequence from the first to ten paths, performing two C-shape turns and a straight movement at each headland. The results of the driving test showed that the reference path was followed with acceptable tracking accuracy when using the autonomous tractor to perform tillage tasks in the field. The RMS lateral errors measured at each straight line ranged from 3.8 to 12.8 cm. The traveling behaviors of the tractor at each path, including straight and curved lines, were similar.
Fig. 6 The actual path of the autonomous tractor obtained from a field test