Changying Charlie Li, Ph.D. Associate Professor University of - - PowerPoint PPT Presentation

▶
changying charlie li ph d associate professor university
SMART_READER_LITE
LIVE PREVIEW

Changying Charlie Li, Ph.D. Associate Professor University of - - PowerPoint PPT Presentation

Jun 05, 2023 274 likes •749 views

Changying Charlie Li, Ph.D. Associate Professor University of Georgia AgRa Webinar October 24, 2013 E-nose Fluorescence imaging of plants and cotton trash Multi- sensor platform Berry Impact Recording Device // Monte Carlo algorith

slide-1
SLIDE 1

Changying “Charlie” Li, Ph.D. Associate Professor University of Georgia AgRa Webinar October 24, 2013

slide-2
SLIDE 2

2

0 1 0 0 1 0 1 0 0 1 0 0 1 1 1 0 1 1 0 1 1 0 1 1 0 0 0 1 0 0 1 0 1 1 0 1 0 0 1 1 1 1 0 0 0 1 1 1 1 0 1 1 0 1 0 0 1 0 1 0 1 1 0 0 0 0 0 0 1 0 0 1 1 0 0 0 0 1 0 1 0 0 1 0 1 1 0 1 0 0 1 1 1 0 0 1 0 1 1 1 1 0 0 0 0 1 0 0 1 0 1 0 0 1 0 0 1 // Monte Carlo algorith main() { char str[STRLEN]; sprintf(str,”Parallel printf(“%s\n”, str); Parallel_Run(); Results_Processing(); ...........

Berry Impact Recording Device Multi- sensor platform E-nose

Fluorescence imaging of plants and cotton trash

slide-3
SLIDE 3

3

Intelligence: learning, planning, navigation Mobility and manipulation Sensing and perceptions

slide-4
SLIDE 4
  • Hyperspectral imaging for onion quality

inspection

  • Electronic nose for rotten onion detection in

storage

  • Berry Impact Recording Device for blueberry

mechanical harvester improvement

4

slide-5
SLIDE 5
  • Onion grading robot
  • Developing a nose for robots
  • A BerryBot to diagnose harvesters

5

slide-6
SLIDE 6

6

Haihua Wang (former Ph.D. student)

Wang, H., C. Li, and M. Wang. 2013. Quantitative determination of onion internal quality using hyperspectral imaging with reflectance, interactance, and transmittance modes. Transactions of

  • ASABE. 56(4): 1623-1635.

Onion grading robot

slide-7
SLIDE 7

Advancing Onion Postharvest Handling Efficiency and Sustainability by Automated Sorting, Disease Control, and Waste Stream Management

7

  • I. SCRI Onion Postharvest Projects

 USDA competitive grant: Specialty Crops Research

Initiative ($774,581)

 Multi-state, comprehensive 4-year research/ extension

project to take onion postharvest handling to next level

slide-8
SLIDE 8

 Onion is the largest vegetable in GA and third

largest in the U.S. ($1 billion)

 13% of the total onion production in the U.S.

goes to dehydration and processed market

 Internal quality (e.g., dry matter) is important  Nondestructive sensing methods are not

available for onion industry.

8

http://www.baldorfood.com

slide-9
SLIDE 9

Easily get fatigued

Fail to detect internal defects and latent fungal diseases

Labor intensive and high cost (50%)

Unable to evaluate internal quality properties

9

slide-10
SLIDE 10

Refractometer (SSC) Magness‐Taylor testing platform (Firmness) Oven (DM)

10

slide-11
SLIDE 11

11

Spectroscopy sugar content prediction for apples, cantaloupes, prune, papaya, tomatoes Birth et al. 1985: onion Spectral imaging

  • external defects

detection

  • diffuse reflectance
  • none for onion

Wavelength (λ) Reflectance Pixel spectra at (x,y)

slide-12
SLIDE 12

12

slide-13
SLIDE 13

13

slide-14
SLIDE 14

14

1

2 2 2 2 2

   a Z b Y X

 cos R R 

2 2 2

cos z y x y P N P N       ] ) ( , ) ( ) ( , ) ( ) ( [ 2 1

2 2 2 2 2

d x j z i b x j a z i d b x j a P P N         

slide-15
SLIDE 15

15

slide-16
SLIDE 16

16

Reflection Interaction Transmission

slide-17
SLIDE 17

17

slide-18
SLIDE 18

18

slide-19
SLIDE 19
  • This

study proved efficacy

  • f

hyperspectral imaging for onion internal quality prediction.

  • Interactance mode can be used to reliably predict

SSC and DM of onions.

  • Next step: implement interactance in packing

lines

19

slide-20
SLIDE 20

20

Tharun Konduru (former M.S. student)

Let the robot have a nose

slide-21
SLIDE 21

 Annual production and storage losses in onion as a

result of diseases can reach 50% or more;

 Botrytis neck rot (caused by the fungus Botrytis allii) and

sour skin (caused by the bacterium Burkholderia cepacia) are most serious threats.

Botrytis neck rot Sour skin

21

slide-22
SLIDE 22

22

slide-23
SLIDE 23

 Develop a customized and low cost gas sensor

array (E-nose)

  • Mechanical
  • Electronic
  • Software

 Test the sensor for sour skin disease detection

in onions

23

slide-24
SLIDE 24

 7 MOS sensors + Temp + RH sensors

24

Pump Valve Gas sensors Teflon chamber Temp/RH sensors Gas inlet Exhaust Clean air inlet

slide-25
SLIDE 25

25

slide-26
SLIDE 26

26

slide-27
SLIDE 27

27

slide-28
SLIDE 28

28

slide-29
SLIDE 29

29

  • Sample preparation: jumbo yellow onions were bought

in local store; surface sterilized

  • Inoculation and incubation: Burkholderia cepacia, strain Bc

98-4; 1mL of bacterial inoculum was injected on two

  • pposite sides of the neck region of the onion (~30mm

deep) X 8 X 8

slide-30
SLIDE 30

30

Batch 1 Batch 2 Control Diseased Control Diseased 3rd dai 16 16 24 24 4th dai 16 16 24 24 5th dai 24 24 23 23 6th dai 24 24 24 24 7th dai 24 24 24 24 Total 104 104 119 119 Total = 446

slide-31
SLIDE 31

31

Diseased Healthy

slide-32
SLIDE 32

32

slide-33
SLIDE 33

33

LDA SVM

All S2,S3,S4 S5,S6,S7 S2,S3,S4 S6,S7 All S2,S3,S4, S5,S6,S7 S2,S3,S4 S6,S7 B1->B2 81.58 56.3 43.15 85.26 81.05 75.78 Average 88.24 86.26 87.63 91.8 92.34 92.36 Leave-1-out 89.89 88.5 88.52 92.35 91.53 92.62

  • SVM is better than LDA
  • Two cross validation methods were better than B1->B2
  • Sensor reduction to 5 could be achievable.
slide-34
SLIDE 34

 A low cost gas sensor array was successfully

developed with an automated gas delivery system and data acquisition features

 Validation tests showed that the device can

differentiate sour skin infected onions from healthy

  • nions starting from four days after inoculation.

 The sensor has the potential to be used for onion

disease detection in storage.

34

slide-35
SLIDE 35

 Rot onion tracing in a large storage room

Concentration mg/kg 12.5 25 37.5 50 62 75 87.5 100

35

slide-36
SLIDE 36

36

Development of a Smart Blueberry

Pengcheng Yu (Former M.S. student) Funded by SCRI blueberry mechanical harvest project

BerryBot to diagnose machine harvesters

slide-37
SLIDE 37

Blueberry Mechanical Harvester

37

Rotary harvester

slide-38
SLIDE 38

38

Berry Impact Record Device (BIRD)

(1) BIRD Sensor node (2) BIRD Interface box (3) PC‐BIRD Software (4) DC Power supply for the interface box

Overall goal: to develop an “instrumented sphere” sensor to measure impacts, identify sources of bruising and optimize mechanical harvesters

slide-39
SLIDE 39

39

BIRD Sensor

slide-40
SLIDE 40

40

BIRD at Work

slide-41
SLIDE 41

Blueberry Mechanical Harvest Field Test

slide-42
SLIDE 42

Real Time Impacts (Rotary)

42

Time (s) Impact (g) 100 200 300 400 500 600

Phase 1 Phase 2 Phase 3 Phase 4

0.7 2.2 6.9 7.3 4 6

Time (s) 0.696 0.698 0.700 0.702 0.704 0.706 0.708 Impact (g) 100 200 300 400 500

slide-43
SLIDE 43

43

Sensor design met design criteria:

Size (25.4 mm)

Frequency (3 kHz)

Memory (1 MB)

Battery (2.5 h)

Sensing range (500g)

Accuracy (0.53%)

Cost ($350)

Field test:

Quantitatively measures impacts during mechanical harvesting (rotary)

Identified critical control points

slide-44
SLIDE 44

 Collaborators  Students, postdocs, visiting scholar, technician.

46

slide-45
SLIDE 45

45

slide-46
SLIDE 46

Thank you!

46

cyli@uga.edu http://sensinglab.engr.uga.edu