CITRUS MECHANICAL HARVESTING:MACHINE ENHANCEMENT ANDMACHINE ENHANCEMENT AND

IMPROVEMENTIMPROVEMENT

Reza EhsaniAssistant ProfessorAssistant Professor

Citrus Research & Education Center

Machine Enhancement and Improvement

• Yield monitoring• Yield monitoring• Developing a rugged fruit mass flow sensor

• Reducing tree injuries through improved• Reducing tree injuries through improved canopy shaking mechanisms• Study the parameters that effect fruit removal or causesStudy the parameters that effect fruit removal or causes

tree injuries

• Variable rate shakingg• Measuring canopy size and volume

• Developing a control system to adjust the optimal h ki tshaking parameters

Yield Monitoring System for Citrus Mechanical Harvesting MachinesMechanical Harvesting Machines

• Yield monitor can be used to:• Quantify the yield variability within or between a

citrus orchard block• Prevent overloading or under loading of citrus• Prevent overloading or under-loading of citrus

transportation truck

• Yield monitoring system for citrus mechanical harvesting machines

Si l• Simple• Easy to install and operate• Easy to maintain and calibrateEasy to maintain and calibrate

• Rugged and durable• Cost-effective

Components of a Yield Monitor for Citrus Mechanical Harvester

GPSTilt Sensor

L d / V l Signal Control and Conditioning Unit

Display UnitLoad / Volume Sensor

Ground Speed B lt S dGround Speed Sensor

Belt Speed Sensor

Mass Flow SensorSYSTEM COMPONENT/s-Two photo interruption layers

- 20 laser beams per layer

ADVANTAGE-Absence of defocus problem

20 laser beams per layer- PIC microcontroller

Absence of defocus problem-Long life calibrationDISADVANTAGE- erroneous signals due to vibrations.itpcΔ erroneous signals due to vibrations.

3 4 b

Mass flow sensor with 40 laser beams

itpsΔ

1 2

⎟⎟⎠

⎞⎜⎜⎝

⎛+

ΔΔ

=

⎟⎟⎟⎟⎟⎞

⎜⎜⎜⎜⎜⎛

+Δ

Δ=

⎟⎟⎟⎟⎟⎞

⎜⎜⎜⎜⎜⎛

+Δ

Δ=

∑

∑

∑∑

∑∑=== 111ˆ 111

i

iTN

N

ii

TNN

N

ii

N

ii

T tpstpcN

tps

tpcN

tpsv

tpcvNN

T

T

TT

TT

CONSTRAINT

itpcΔ

2

1 3

4

1+Δ itpcitpsΔ⎠⎝

⎟⎟⎠

⎜⎜⎝

Δ⎟⎟⎠

⎜⎜⎝

Δ ∑∑∑=== 111

i

ii

ii

ii

ptpstpsv

Poisson FlowitfΔ 1+Δ itf

Dual interruption timing mechanism

Impact PlateSYSTEM COMPONENT/sFour (4) Load Cells

ADVANTAGE

- Four (4) Load Cells- Tern Controller, GPS

Plate displacement-Rugged and easy to handle-Easy to install and maintain

IMPACT FORCE

∫ ∫+= max max

0 0

2

21 δ α

αδ FdFdMv

IMPACT FORCE

PLATE DISPLACEMENT

( ) ( )( ) ( )22

2

fNfSfS

f+

=Φ

PLATE DISPLACEMENT

( ) ( )( )⎧ >

⎩⎨⎧

−>−

=.

, Clumps

TxforClumps

otherwiseTxTxforx-T αβ

FILTER

⎩⎨⎧ >

=.0,

Weight Pr.otherwise

TxforClumps

WEIGHT

LABORATORY & FIELD TEST

Field Calibration

y = 1.3567x + 411.3110000

12000

y 1.3567x 411.31R² = 0.9725

8000

Wei

ght (

lb)

4000

6000

True

Net

W

2000

0

0 1000 2000 3000 4000 5000 6000 7000 8000 9000

Computed Weight (lb)

Yield Monitoring System in Action

14500Top

13500

14000Bottom

13000

13500

12500

eld

(lb)

11500

12000Yie

11000

10000

10500

Treat. 1 Treat. 2 Treat. 3

Treatment

Reducing Tree Injuries & Improving Fruit g j p gRemoval

Ti d i i bl

Force Distribution in the Tree Canopy during Mechanical Harvesting

Tine design variablesForward SpeedAmplitudeF

X(t)

X(t)

FrequencyInclination (α)

X(t)φ

C K

φ

C, K

Branch

Stem

ϕC, K

W

F

ϕ

WF

High

Shaking forceHigh

Medium

αLow

Inclination

Reducing Tree Injuries & Improving Fruit g j p gRemoval

• Study the force distribution in a canopy of citrus trees and determine the best shaking parameters fortrees and determine the best shaking parameters for each type and size of tree • Study the force distribution in an individual branch in the tree whenStudy the force distribution in an individual branch in the tree when

harvested by a canopy shaker

• Study the effects of frequency of shaker, angle of tines and tree size on the f di t ib tiforce distribution

• Variable rate shakingM i i d l• Measuring canopy size and volume

• Developing a control system to adjust the optimal shaking parameters

Independent Variables

Tree size• Tree size• Angle of tines• Frequency of the canopy shaker• Frequency of the canopy shaker

Independent Variable Levels ValuesIndependent Variable Levels Values

1 Tree Size (2 Levels)

1 Large(2 Levels)

2 Medium

2 Angle of Tines (3 Levels)

1 5°

2 20°

3 35°

3 Frequency of canopy shaker

1 200 cpmshaker

(3 Levels) 2 250 cpm

3 300 cpm

MethodologyMethodology

Instrumenting the branch:Instrumenting the branch:

• Instrumenting the selected branch starts from the base of the branch where it is joined with the main trunk.j

• A sensor is placed at the base.

• From the base of the branch, the sensors are placed at distances of multiples of 50 cm.

The last sensor is placed such that the diameter of the branch at that• The last sensor is placed such that the diameter of the branch at that location is about 15 cm.

Images of a Tested Tree

Videos of ExperimentVideos of Experiment

PRELIMINARY RESULTSPRELIMINARY RESULTSLarge Tree

Preliminary ResultsMedium Tree

Variable Rate Shaking of Trees A di Si b C Sh kAccording to Size by Canopy Shakers

Laserscanner

Control unit Motor

Control unit

Computer

LMSprogram

Controlprogram

Voltage = f(freq.)

Voltagevariable actuator

Tree heightTree volume

Tree shape

Shakingfreq. VoltageLaser

scannerShaking

mechanism

Feedback signal

Tree Canopy Volume MeasurementL G G i GPSLaser scanner(SICK LMS200)

Gyro sensor(VG440-CA)

Garmin GPS(GPS18-5Hz)

Notebook t

LabVIEW

computer(CPU: 2.0 GHz)

xk= TD - dk⋅sin(θk)yk= SH + dk⋅cos(θk)

LMS. dk

θkpk(xk,yk)(xk,yk)(-xk,yk) VVtreeofVolume

StAVsliceofVolumen

ii

ii

=

⋅Δ⋅=

∑=

)(

)(

1

yk= SH + dk⋅cos(θk)

LMS..

SH polygonofareaAspeedvehicleS

cyclepertimescanningtwhere

i ===Δ,

TD(0,0)

x(0,0)

Tree Canopy Volume MeasurementPolygon area by raw data

By convex hull method

By hull fit method (P=0.15)

* Roll angle correction

Tree geometric

characteristics

Travel speed of the vehicle= 1.0 (m/sec)

Meas-True Error (%)

Tree Height / Volume

Raw data / Roll angle Correction

Travel speed of the vehicle= 1.0 (m/sec)

(roll angle = 8.8°)

Height (cm)(True=240) -0.991 -0.41

Volume (m3)

Raw data -1.580 -13.99

Convex hull 0.665 5.89

Meas-True Error (%)

Height (cm)(True=240)

Raw data -43.510 -18.13

Roll angle correction -6.770 -2.82

Volume (m3)(True=11.286)

Hull fit

P=0.15 0.010 0.09

P=0.30 0.409 3.62

P=0.60 0.585 5.18

Volume (m3)(True=11.286)

Hull fit(P=0.15)

Raw data -0.610 -5.40

Roll anglecorrection -0.060 -0.53

Thank YouThank YouAny Questions?