Update: Automation for Bare Root Ornamental and Fruit Tree Inventory
1. Update: Automation for Bare Root
Ornamental and Fruit Tree
Inventory
J.S. Owen Jr., W. Shi, H.M. Stoven,
S. Singh, K. Lewis, S. Doane, M. Bergerman
2. Caliper, Counter and GPS tracking
- Year 3 report
Review of goals
What is new with Caliper
What is new with Counter
Location tracking with inexpensive GPS
Caliper, Counter and GPS field testing and
demo in WA, OR and CA
Outreach
Barriers to adoption / Economics analysis
3. 3-year goals
Activities Deliverables Success Criteria
1a. Redesign and reimplement 1a. New on-the-fly caliper 1. Caliper measurements within
on-the-fly caliper device 1b. Report on the comparison of 1/8” (3 ) of staked bare root
1b. Design and construct 3-point caliper device data to actual in fruit, flowering, and shade trees
hitch for caliper in-field testing WA and OR for bare-root at 6” (±2”) above the graft union
fruit, flowering, and shade trees or soil travelling 3 mph or
greater using a 3-point hitch
2. Continue refinement and 2. Report on the comparison of 2a. Count trees greater than ¼”
integration of counter device to counter device data to actual in (6 mm) caliper with 96%
count bare-root fruit, flowering, WA and OR for bare-root accuracy at 3 mph or greater
and shade trees before harvest; fruit, flowering, and shade trees 2b. Count trees greater that ½”
all in-field tests conducted using (13 mm) caliper with 98%
existing, on-farm equipment accuracy at 5 mph or greater
while conducting simultaneous
production-related tasks
3. Conduct field tests in WA and 3. Reports of tests conducted 3. Repeated trials, day and
OR at night and day using 3- night, at OR and WA nurseries
point hitch implemented and reported
4. Identify hardware or software 4. Low-cost counter device with 4. Counter device sale price <
needed for integration of integrated electronics and $2,000
device(s) that lead to successful processing
commercialization
4. Caliper - Field
New caliper design and built to overcome
stake
Design and fabricate 3-point hitch for caliper
in-field testing
Caliper measurements need to be
determined at:
distance of 10 to 20” from tree
6” (±2”) above the graft union or soil travelling
travelling 3 mph
Caliper measurements within 1/8” of staked
bare root fruit, flowering, and shade trees
5. Counter - Field
Purchase 2 new counters operated by netbook
Imputable size info to increase accuracy?
Design and fabricate dual mount for counters on:
Tractor
ATV
Count trees greater than ¼”
(6 mm) caliper with 96% accuracy at 3 mph or greater
Count trees greater that ½” (13 mm) caliper with 98%
accuracy at 5 mph or greater
6. Caliper Hardware
A new caliper
Low power, sharper lasers
Reduced laser baseline
New steel casing
Improved existing caliper
Sharper lasers
More accurate calibration
Wide field of view
7. Caliper Software
Completely rewrote to improve
efficiency – can process caliper
estimation at 60 times per
second (60 Hz from 7 Hz last
year)
Added process to handle trees
with stakes
Created 2 versions – field and
warehouse calipers
Added Graphic User Interface
(GUI) and features based on
grower feedback
8. Warehouse Caliper GUI
User can specify info such as location, tree, grading and
bin capacity
Logs data, show running status
Shows bin count and alerts user when bin is full
9. Field Caliper GUI
User can specify if stakes are used and diameter of stake
Reports optimum measuring distance range and maximum
allowed traveling speed to ensure adequate estimates per
tree (approx. 12)
10. Caliper field tests
J Frank Schmidt Nursery, OR
Caliper mounted at the front of tractor
11. Caliper field tests at J Frank Schmidt
nursery on Aug 24-25 2011
Caliper testing scenario:
Testing on Red Pointe Maple, 2 years old
100 trees measured, total length 112 feet
~12 inch spacing in between trees
Trees staked with metal stakes (0.25 in)
Runs at 1 mph, 2 mph and 3 mph
Manually measured the caliper of 100 trees as
Ground Truth
13. Caliper field tests at J Frank Schmidt
nursery on Aug 24-25 2011
Findings:
Increased number of hits but this configuration is
more sensitive to errors in keeping sensor at
constant distance from trees.
Still not able to satisfactorally remove the
stake from calculation of diameter.
14. Caliper field tests at J Frank Schmidt
nursery on Aug 24-25 2011
Example:
4 of 23 measurements on one tree
Stake does not affect any of 23 estimates
stake
15. Examples of Failures
3 images of set of 22 estimates
on another tree
Stake affects on 6 estimates
Having large # of estimates is
important – can compensate for
errors in detecting stakes
16. Caliper field tests at J Frank Schmidt
nursery on Aug 24-25 2011
Limitation for staked trees
Need to know the stake’s caliper
Stake’s caliper must be significant difference than average tree
caliper (ratio of stake caliper to tree caliper)
Left figure shows stakes deemed acceptable, right figure shows
stakes that are not viable.
Remain unable to measure staked trees regardless.
21. Counter Hardware
2011 Improvements
Added optics to diffuse laser
beam (by 5X) to get more
hits on the trees
Optics dramatically improve
performance – accuracy &
speed of travel possible
22. Counter Software
Added Graphic User Interface to include user
input, pausing and saving data
24. J Frank Schmidt
Staked 2 year old red
pointe maple
Single row of 708 trees
½ - 1 ¼ caliper
12 inch spacing
Run# Speed #Trees Count Err % Log file
1 3mph 708 681 3.8% runCounterAt.2011-08-25.18:18:59
2 3mph 708 698 1.4% runCounterAt.2011-08-25.18:34:24
3 3mph 708 639 9.7% runCounterAt.2011-08-25.18:56:13
4 3mph 708 696 1.7% runCounterAt.2011-08-25.19:04:42
5 3mph 708 625 11.7% runCounterAt.2011-08-25.19:13:15
6 3mph 708 716 1.1% runCounterAt.2011-08-25.19:19:49
25. J Frank Schmidt
Staked 2 year old red
pointe maple
Single row of 708 trees
½ - 1 ¼ caliper
12 inch spacing
Run# Speed #Trees Count Err % Log file
1 3mph 708 681 3.8% runCounterAt.2011-08-25.18:18:59
2 3mph 708 698 1.4% runCounterAt.2011-08-25.18:34:24
3 3mph 708 639 9.7% runCounterAt.2011-08-25.18:56:13
4 3mph 708 696 1.7% runCounterAt.2011-08-25.19:04:42
5 3mph 708 625 11.7% runCounterAt.2011-08-25.19:13:15
6 3mph 708 716 1.1% runCounterAt.2011-08-25.19:19:49
26.
27. Dave Wilson Nursery, CA
All almond trees on 5” spacing in row ¼” to 5/16” caliper
Nonpareil Almond on Nemaguard rootstock (1/4”)
Wood Colony Almond on Nemaguard (5/16”)
Honey Royale Almond on Nemaguard (5/16”)
Speed Run Run Run Run Run Run Run Run Mean
Err%
mph 1 2 3 4 5 6 7 8
2.5 52 50 53 51 52 - - - 3.2
3.0 51 52 51 53 48 52 49 48 1.0
4.0 52 51 52 47 - - - - 1.0
5.0 54 47 50 51 49 48 - - 0.4
28.
29. GPS tests for localization
GPS data recording on rows of tall trees with canopy
which potentially blocks the view of GPS satellites
Device:
○ U-blox 6T
30. Location tracking with low-cost GPS
GPS data is logged with
cheap(<$300)GPS unit while conducting
field tasks, such as running caliper or
counter
Data is post-processed with correction
data from US Govt. the day after
collection
Field tested in spring (WA) and fall (OR)
31.
32. Washington Tree Fruit Nursery
Spring 2010
Tractor track:
raw data in
red
Processed
data in green.
33. J Frank Schmidt Nursery
Fall, 2011
Tractor track: raw
data in red
Processed data in
green.
Processed data
jumps by 1m
occasionally due to
canopy
34. J Frank Schmidt Nursery
Fall, 2011
Tractor’s speed plot: during travel at
~3.0 mph
36. Outreach
Numerous stakeholder meetings
Presented to Northwest Nursery Improvement Institute (NNII)
Demonstration and presentation at Farwest Show
Research presented to ANLA on ‘Kick the Dirt’ Educational Sessions
Led education Farwest session on Automation and Technology that
included presentation from Clark Seavert on value proposition.
39. Jim Owen’s New Contact
Info
Virginia Polytechnic Institute and State University
Hampton Roads Agricultural Research and Extension Center
1444 Diamond Springs Road, Virginia Beach, VA 23455
Tel: 757.363.3904
Lab: 757.363.3906
Fax: 757.363.3950
Email: nsy.prod@vt.edu
Skype/Google: nsy.prod.vt
Web: www.arec.vaes.vt.edu/hampton-roads/