engineering - carnegie mellon school of computer...
TRANSCRIPT
Paul Heinemann, Penn State Dept. of Ag & Biol. Engineering
Tara Baugher, Penn State Extension
Advanced Technologies for Thinning of Tree
Fruit -- Engineering Solutions Workshop, CMU
11-27-12
Partners involved:
• Penn State
• WSU
• UC-Davis
• Clemson
• U Maryland
• U Illinois
• CMU (selective thinning)
• USDA
• Extension
• Grower partners
• Other industry partners
Problem Statement
Fruit thinning is a labor-intensive and expensive task; time and crew availability limitations often result in orchards not being thinned in an optimal way for maximum fruit quality
Why is fruit thinned?
Crowding results in low quality, small fruit
Thinning results in high quality, large fruit
Technology – mechanized thinning
Two levels of technology are being used for fruit thinning mechanization:
Non-selective
Selective
String thinner positioning – making the work easier for the operator
Two-tiered approach to controlled positioning
• Joystick control
• Sensor control
Joystick control
This is an interim step for providing easier manual positioning control
Joystick for Tilt and Offset
E- Stop
Autonomy Controls
LED Indicator Lights
Controller adjusts the spindle accordingly
Two degrees of motion: angle of spindle, lateral position of spindle
hydraulics and controls added
components: Proportional
Control
Valves
Tilt
Cylinder Lateral
Offset
Cylinder
Flow
Divider
Ultrasonic Sensing
• Arduino Microcontroller
– $65
• 4 Sonar Range Finders
– Maxbotix LV-MaxSonar-WR1
– $100
• Low Cost GPS
– USGlobalSat EM-406A GPS Module
– $60
Sensor Mast
z
y
x
Best Fit Line
Orchard Row
Sensor Distance Readings (dashed lines)
Ultrasonically Sensed Position
Selective Thinning
Select blossoms/small fruit to be removed for higher precision in thinning using machinery
: Real-time acquisition
25
Real-time Industrial Cameras
Texture Illumination (reverse “view camera”)
Strobe
Credit: UC Davis
Building the 3D Map of all Trees in a row from the Individual Sequences
VectorNAV (yaw,pitch, roll)
GPS (x,y,z)
R-camera
L-camera
Θ Pattern strobe (or floodlight)
Lidar
Credit: UC Davis
Real-time 3D Tree and Blossom Mapping System
27
Lidar
Right Camera
Left Camera
Pattern Strobe
GPS
Tilt Sensor
Credit: UC Davis
End-Effector
2nd Generation Brush End-Effector
Improvements
Electric powered
Shorter Brushes
Softer Brushes
End-Effector
1st Generation Hand Gripper End-Effector
Follow-Up Hand Thinning $100 to $400/acre savings
Hrs/ acre
Treatments with different letters are significantly different at P=0.05
Molded Strings - 2 Rows - No Gaps - 2 mph
y = 0.528x - 38.833
R2 = 0.9872
y = 0.588x - 46.533
R2 = 0.9815
y = 0.788x - 82
R2 = 1
y = 0.614x - 36.883
R2 = 0.9308
y = 0.552x - 49
R2 = 0.9845
y = 0.588x - 53.967
R2 = 0.9278
y = 0.436x - 47.7
R2 = 1
0
10
20
30
40
50
60
70
80
90
100
100 125 150 175 200 225 250
Spindle Rotation (rpm)
Flo
we
r R
em
ov
al
(%)
California
200 rpm, 1.8mph220 rpm, 1.8 mph
Control
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Small
Medium
LargePe
rce
nt P
rod
uct
ion
Fruit Size Distribution from Three Thinning Treatments in Nesstar Peaches
South Carolina
Drum Shaker Trials - PA
• String thinner at bloom/drum shaker for green fruit thinning an effective combination
• New drum shaker at green fruit stage more consistent than previous citrus shaker prototypes – Smaller rods, tilting drums, front or side-mount
• Tree damage still a problem
Thinning results from 2011 (controlled positioning)
Upper canopy Lower canopy Scaffold
Manually operated joystick 62.8 a 63.5 a 63.1 a
Ultrasonic controlled positioning 49.7 a 35.4 b 43.7 b
Laser controlled positioning 61.2 a 51.9 a 58.7 a
Upper canopy Lower canopy Scaffold
Manually operated joystick 3.9 a 1.3 a 3.6 a
Ultrasonic controlled positioning 4.7 a 2.7 b 5.8 a
Laser controlled positioning 4.7 a 1.9 a 4.8 a
% blossoms thinned
Blossom density, flowers/cm2 cross-sectional area
Challenges and Limitations
• Non-selective thinners work well with stone fruit, but major challenges with pome fruit
• Selective thinning is an extremely complex task • Locating branches, blossoms, 2-D and 3-D • Mobility of robotic in field • Cost of robotic approach • Speed of activity
Net Economic Impact
Processing Peach Plantings
• $236 to $1490/acre
Fresh Peach Plantings
• $185 to $934/acre
Labor savings and yield increases (due to larger fruit size) were of greater importance for canning peach growers
Increased fruit size had a greater positive impact for fresh market producers
Blossom Thinned with String Thinner
Hand Thinned
Technology Adoption Case Study Interviews
11 growers who had thinned a total of 154 acres
Employees liked thinner - saved time/minimized ladder use
Hand thinning completed earlier - more timely work in other crops
100% said hand thinning time reduced 80% said fruit were larger