AutoTrac – Accuracy of a RTK DGPS based Autonomous Vehicle Guidance System under
Field Conditions
M. Ehrl,W.V. Stempfhuber, M. Demmel, M. Kainz, H. Auernhammer
Centre of Life Sciences WeihenstephanDepartment of Bio Resources and Land Use Technology
Crop Production Engineering
Ehrl / Auernhammer, October 8, 2004 Slide 1© 2004
Crop Production Engineering
Automation Technology for Off-road Equipment (ATOE)Kyoto (Japan)
October 7-8, 2004
Outline
1. Objective
2. AGRO NAV system setup
3. Reference measuring system setup
4. Equipment verification
5. Results
Crop Production Engineering
Ehrl / Auernhammer, October 8, 2004 Slide 2© 2004
6. Conclusions
Subject Matter
• Since about two years, autonomous guidance systems are commercially available for agricultural machines
• The technology is generally based on GPS with Diffe rential Correction (DGPS) or Real Time Kinematic GPS (RTK D GPS)
• Inertial guidance technology is often used for dead reckoning and roll and pitch compensation
Crop Production Engineering
Ehrl / Auernhammer, October 8, 2004 Slide 3© 2004
Until now, only limited information is available on the accuracy in practical use under typical and often d ifficult agricultural field conditions (side slope, wheel sl ip)
Subject of Investigation
• The first commercial system for agricultural machin es was AGRO NAV®, developed by GEO TEC electronics GmbH in Germany, accuracy ± 100 mm
• The tested system was implemented on a standard tractor (MF 4255) and utilized RTK DGPS and an Inertial Measurement Unit
Crop Production Engineering
Ehrl / Auernhammer, October 8, 2004 Slide 4© 2004
Measurement Unit (IMU) for navigation
Schematic System Setup of AGRO NAV ®
CAN 1 (GEO TEC proprietary protocol)
(DIN 9684 / ISO 11783 protocol)
CAN 2 - Implement Bus
AGRO NAV PLAN (PC Software) - Mission Planning - Documentation - Task Management - Data Interchange
PCMCIA / USB Port
Navigation Controller / User Interface (GT 2000, Embedded PC with OS
Embedded Windows NT 4.0) - Kalman Filter Algorithms - Navigation Algorithms - Task Control - GPS / IMU processing - Documentation - Virtual Terminal (LBS / ISOBUS ) - ...
Vehicle Controller (ESX - STW, 16-bit embedded
Controller) - Steering - Cruise Control - Engine Control - Brakes - ...
System Monitoring Unit (all functionality hardware implemented)
Inertial Measurement Unit (IMU) - 1x Fiber Optic Gyro (FOG) - 2x accelerometers
(DIN 9684 / ISO 11783 protocol)
(bidirectional)
RS 485 Command
Proprietary Parallel Port (GEO TEC protocol)
RTK DGPS Ashtech Z-Eurocard - 10 Hz position output - integrated radio link
(unidirectional)
RS 485 Data
(hw handshake)
RS 232
proportional valve (steering control)
Crop Production Engineering
Ehrl / Auernhammer, October 8, 2004 Slide 5© 2004
- priority control / system enable - safety control - I/O port processing
GPS antenna radio link
Backup Battery Buffer - avoids interruptions of the energy supply
proportional valve (steering control) steering angle sensors seat pressure sensor gearbox monitoring throttle actuator electro-hydraulic brake system emergency-stop-button radar odometer system enabling ...
I/O Ports
Reference Measuring System
• Needs to be a completely independent system which d oes not influence the system environment
• The degree of accuracy needs to be at least three t imes better than the desired resolution
• Capable for outdoor measurements in a rough and non -uniform environment
Crop Production Engineering
Ehrl / Auernhammer, October 8, 2004 Slide 6© 2004
The deployed position reference system consisted of two geodetic RTK DGPS receivers (Leica SR530, 10 Hz, rov er mode) and one base station (Leica SR530, 10 Hz, base stati on mode)
Antenna Alignment
• The accurately defined alignment of the antennas is important for data evaluation
• Position deviations in all three dimensions and inc linations
-2.5
-2
-1.5
-1
-0.5
0
Z -
Axi
s
Rover 1
Rover 2
AGRO NAV
Distance Root Point to AGRO NAV = 270 cm
Distance Rover 1 to Rover 2 = 130.5 cm
AntennaAGRO NAV
Root Point(270 cm lower)
• Position deviations in all three dimensions and inc linations can be measured
Crop Production Engineering
Ehrl / Auernhammer, October 8, 2004 Slide 7© 2004
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0
0.5
1 -3
-2.5
X - Axis
Root Point
AGRO NAV = 270 cm
Y - Axis AntennaRover 1
AntennaRover 2
Equipment Verification
• To verify the equipment, tests on a tarred road hav e been carried out
By means of AGRO NAV PLAN ®, a job consisting of a • By means of AGRO NAV PLAN ®, a job consisting of a straight line of 180 m length was planned and trans ferred to the navigation controller
- surface with single cross grooves
- downward slope of about 4.5 percent
• Condition of the tarred road:
Crop Production Engineering
Ehrl / Auernhammer, October 8, 2004 Slide 8© 2004
- sidewise slope changing between 0.5 and 3.5 percent
• Several runs with speeds of 2.0, 4.0, 6.0 and 12 km /h speed were conducted
Measured Parameters
• Cross Track Error (XTE) - most significant parameter , defined as the horizontal distance of the navigational point (center of rear axle) normal to the planned position (set poin t)rear axle) normal to the planned position (set poin t)
• Inclination values of AGRO NAV IMU (roll and pitch) in comparison to the calculated values of the referenc e measuring system
z
yaw
Crop Production Engineering
Ehrl / Auernhammer, October 8, 2004 Slide 9© 2004
roll
pitch
x y
Spatial Relationships
155.6
155.8
m]
Position AGRO NAV (system output)
Position Rover 2
155.0
155.2
155.4
Nor
thin
g (5
3630
00)
[m
Calculated Position
Planned Position
Position Rover 1 XTE (Cross track error)
Crop Production Engineering
Ehrl / Auernhammer, October 8, 2004 Slide 10© 2004
363.0 361.0 361.5 362.0 362.5
154.6
154.8
Easting (4477000) [m]
Planned Path Rover 1 Rover 2 AGRO NAV
Results: Inclination Values of AGRO NAV IMUat 2.0 km/h speed
4
5
-3
-2
-1
0
1
2
3
Ang
le [°
]
Crop Production Engineering
Ehrl / Auernhammer, October 8, 2004 Slide 11© 2004
-5
-4
-3
0 20 40 60 80 100 120 140 160 180 Driving Route [m]
IMU Roll IMU Pitch
Results: Calculated Roll values of Reference System at 2.0 km/h speed
5
-1
0
1
2
3
4
5
Ang
le [°
]
Reference System Roll
Crop Production Engineering
Ehrl / Auernhammer, October 8, 2004 Slide 12© 2004
-2 0 20 40 60 80 100 120 140 160 180
Driving Route [m]
Comparision:Measured inclination values of AGRO NAV IMU
and calculated inclination values (at 2.0 km/h speed)
4
5
0
1
2
3
4
Ang
le [°
]
3
4
5
Crop Production Engineering
Ehrl / Auernhammer, October 8, 2004 Slide 13© 2004
0
1
2
3
Ang
le [°
]
Results: XTE measured by Reference Systemat 2.0 km/h speed
150
200
-50
0
50
100
150
XT
E [m
m]
Crop Production Engineering
Ehrl / Auernhammer, October 8, 2004 Slide 14© 2004
-150
-100
0 20 40 60 80 100 120 140 160 180 Driving Route [m]
XTE (no Roll correction) XTE (with Roll correction)
Practical Investigations
• MF 4255 tractor unmanned with a rear mounted combin ation of rotary harrow and air-seeder (3.0 m working widt h)
• Conditions: autumn, wet soil, winter wheat seeding, field • Conditions: autumn, wet soil, winter wheat seeding, field with side hill slopes between 0 and 14 percent
transect 2
transect 1
transect 3
-100
0
XT
E [m
m]
Crop Production Engineering
Ehrl / Auernhammer, October 8, 2004 Slide 15© 2004
transect 1
0 25 50 75 100
Distance to field border [m]
-200
Practical Investigations
• MF 4255 tractor unmanned with a rear mounted combin ation of rotary harrow and air-seeder (3.0 m working widt h)
• Conditions: autumn, wet soil, winter wheat seeding, field • Conditions: autumn, wet soil, winter wheat seeding, field with side hill slopes between 0 and 14 percent
transect 2
transect 1
transect 3
-100
0
XT
E [m
m]
Crop Production Engineering
Ehrl / Auernhammer, October 8, 2004 Slide 16© 2004
transect 1
0 25 50 75 100
Distance to field border [m]
-200
Results
• Accuracy showed similar results as on the tarred ro ad
• Wheel slip and downhill drift of the tractor have a lso been properly controlled by the guidance systemproperly controlled by the guidance system
• Work result was not in the expected range of ± 100 m m
• Deviations of up to 240 mm from path to path were c aused by the downhill yawing of the rear of the tractor
• Possible Solutions:
Crop Production Engineering
Ehrl / Auernhammer, October 8, 2004 Slide 17© 2004
• Possible Solutions:- rigid connection of the implement
- implement mounted antenna instead of tractor mounted
- sensing the relative position of implement and tractor (e.g. for trailed implements)
Conclusions
• Reference measuring system has proofed practicabili ty
• Reference system is able to measure inclinations - a ccuracy can be improved by an extended antenna spacingcan be improved by an extended antenna spacing
• AGRO NAV® was in the specified range of ± 100 mm for speeds up to 10 km/h (for 12 km/h in the range of ± 130 mm)
• Deviations on a sloped field up to 240 mm caused by yawing of the implement
• Planning Software was only able to create jobs in 2 D (x-y
Crop Production Engineering
Ehrl / Auernhammer, October 8, 2004 Slide 18© 2004
• Planning Software was only able to create jobs in 2 D (x-y plane) instead of 3D – sloped contours were not cons idered
• The great potential of RTK DGPS based auto guidance technology for agricultural vehicles was clearly de monstrated