Transcript AGRICULTURE: A Field for Development using

1
AGRICULTURE: A Field for
Development using AI Techniques
- Lets Identify the Applications
CS621-Artificial Intelligence Course Seminar
Presented by:
V S K Murthy B (08407403)
Singre Pawan (07305039)
CS621 Course Tutor:
Prof. Pushpak Bhattacharya
2
CS621-Artificial Intelligence Course Seminar
Outline
Talk is divided into two parts:
• Part-I:
▫ Why to choose “field of Agriculture” ?
▫ Identified Areas for enhancing Agriculture sector
▫ Computational Intelligence in Agriculture and Environment
• Part-II:
▫ Intelligent Environment Control for Plant Production
▫ Intelligent Robot in Agriculture
• Conclusion
V S K Murthy and Singre Pavan
3
CS621-Artificial Intelligence Course Seminar
Part-I
V S K Murthy and Singre Pavan
4
CS621-Artificial Intelligence Course Seminar
Why to choose “Field of Agriculture”?
• Sector status in India
▫ Growth of socio-economic sector in India
▫ Means of living for almost 66% of the employed class in India
▫ Acquired 18% of India's GDP
▫ Occupied almost 43% of India's geographical area
• Huge investment made for Irrigation facilities etc. in 11th five year
plan
• Introduction of de-regulation in agriculture sector
▫ Opens competition for agriculture products
▫ Removal of unnecessary restrictions — movement, stocking, and
so on..
▫ Good price to farmer
▫ Substantial technology growth in coming years
V S K Murthy and Singre Pavan
5
CS621-Artificial Intelligence Course Seminar
Why to choose “Field of Agriculture”?
Philosophy
of
Efficiency
Efficiency curves
Different Technologies
Time scale
• Any process growth rates can be linked with efficiency curves
• Due to deregulation, Agriculture has bright future insight
V S K Murthy and Singre Pavan
6
CS621-Artificial Intelligence Course Seminar
Why to choose “Field of Agriculture”?
• Peak in the agricultural sector will again reach in near future
V S K Murthy and Singre Pavan
7
CS621-Artificial Intelligence Course Seminar
Identified Areas for enhancing
Agriculture sector
• Needs monitoring on
▫ Agricultural crop conditions
▫ Weather and climate
▫ Ecosystems
• Decision support for agricultural planning and policy-making
• On the basis of AI interest
▫ Computational Intelligence in Agriculture and the Environment
 Optimizing different types of bio-systems
 Testing and fitting of quantitative models
▫ Intelligent environment control for plant production systems
▫ Intelligent robots in agriculture
▫ An expert geographical information system for land evaluation
▫ Artificial neural network for plant classification using image processing.
▫ Control of green house.
V S K Murthy and Singre Pavan
8
CS621-Artificial Intelligence Course Seminar
Computational Intelligence in
Agriculture and the Environment
V S K Murthy and Singre Pavan
9
CS621-Artificial Intelligence Course Seminar
• Search procedures
▫
▫
▫
▫
Exhaustive techniques (random walk)
Calculus based methods (gradient methods)
Partial knowledge techniques (hill climbing)
Knowledge based techniques (Production rule systems, heuristic
methods)
▫ Stochastic techniques (SA)
▫ Biologically inspired algorithms (GA and immune)
• Problems deal with optimizing bio-systems and fitting
quantitative models require
▫ Refinement or processing using adaptive search procedures
• Bio-system Derived Algorithms (BDAs)
▫ Photosynthetic Algorithm (PA)
▫ Leaf Cellular Automate (LCA)
V S K Murthy and Singre Pavan
10
CS621-Artificial Intelligence Course Seminar
Photo-Synthetic Algorithm
Atmosphere
Light
(Stimulation)
o Any problem that can
be solved by GA can
also be solves by
PS Algorithm
BensonCalvin
Cycle
Oxygen/CO2
concentration
CO2
Reservoir
RuBP
PhotoRespiration
GAP
Copy Good
Next Iteration
Fitness
Discard
Poor
DHAP
(Knowledge
string)
V S K Murthy and Singre Pavan
11
CS621-Artificial Intelligence Course Seminar
Similarities of GA and PA Algorithms
Genetic Algorithm
•
•
•
•
Initial Population
Crossover of Chromosomes
Mutation
Final form of chromosome
(Solution)
Photo-synthetic
Algorithm
• Co2 Concentration
• Benson-calvin cycle (includes
many different
recombinations of molecules)
• Bio-chemical balance
between the BCC and PRS
• DHAP considered as the
Knowledge string (Solution)
Example: In Part-II, Nutrient control set for plant growth has been solved by
PS Algorithm
V S K Murthy and Singre Pavan
12
CS621-Artificial Intelligence Course Seminar
Part-II
V S K Murthy and Singre Pavan
13
CS621-Artificial Intelligence Course Seminar
Intelligent Environment Control For
Plant Production System
V S K Murthy and Singre Pavan
14
CS621-Artificial Intelligence Course Seminar
Why it is required?
• To increase productivity of crops
• Care for special herbal valued plants, environment diverse
plants etc., which in turn increases our export value
• To develop decision making support
V S K Murthy and Singre Pavan
15
CS621-Artificial Intelligence Course Seminar
Hydroponic System
V S K Murthy and Singre Pavan
16
CS621-Artificial Intelligence Course Seminar
V S K Murthy and Singre Pavan
17
CS621-Artificial Intelligence Course Seminar
Plant Growth Optimization Problem
• In plant production, good fruit yield requires an optimal
balance between
▫ Vegetative growth (e.g. root, stem, leaf growth)
▫ Reproductive growth (e.g. flower and fruit growth)
• NNs and GA provides optimal set points of the nutrient
concentration (NC).
• The ratio of total leaf length (TLL) to stem diameter (SD)
defines as a predictor for plant production growth.
V S K Murthy and Singre Pavan
18
CS621-Artificial Intelligence Course Seminar
Optimization Problem
• Let TLL(k)/SD(k) be the time series of TLL/SD as affected by NC(k)
(k=1,......,N; N : final day)
• Seedling stage(1 ≤ k ≤ N ) divided into four steps:
▫ Transplanting
▫ Vegetative growth after transplanting
▫ Flowering of the first truss
▫ Fruit setting for the first truss and flowering for the
second truss.
• Consider the value of nutrient concentration at each step is NC1,
NC2, NC3, NC4 .
{1≤ k ≤ N1L : step1, N1L+1 ≤ k ≤ N2L: step2,
N2L+1 ≤ k ≤ N3L : step3, N3L+1 ≤ k ≤ N : step4}
V S K Murthy and Singre Pavan
19
CS621-Artificial Intelligence Course Seminar
Optimization Problem
• Objective Function :
N
1
TLL( K )
F ( NC ) 

N  N 3 L  1 K  N3 L 1 SD( K )
• Objective Problem
Maximize
Subject to
F(NC)
α1 ≤ NC(k) ≤ α2
V S K Murthy and Singre Pavan
20
CS621-Artificial Intelligence Course Seminar
Neural Networks
V S K Murthy and Singre Pavan
21
CS621-Artificial Intelligence Course Seminar
Genetic Algorithm
V S K Murthy and Singre Pavan
22
CS621-Artificial Intelligence Course Seminar
Procedure of GA
• Step1: The Initial population consisting of several individuals
• Step2 : Several individuals in another population are added to
original population to maintain diversity
• Step3 : Crossover and mutation operations are applied to the
individuals
• Step4 : Fitness values of all individuals are calculated by NN
model
• Step5 : Superior individuals are selected and retained for next
generation
• Step6 : step 2 through 5 are repeated until an arbitrary
condition satisfied
V S K Murthy and Singre Pavan
23
CS621-Artificial Intelligence Course Seminar
Intelligent Robots in Agriculture
V S K Murthy and Singre Pavan
24
CS621-Artificial Intelligence Course Seminar
Strawberry harvesting robot
Source: http://www.lovingthemachine.com/2008/04/farmer-hails-strawberry-picking-robot.html
V S K Murthy and Singre Pavan
25
CS621-Artificial Intelligence Course Seminar
Hortibot robot for weeding
Source: http://www.lovingthemachine.com/2008/04/farmer-hails-weeding.html
V S K Murthy and Singre Pavan
26
CS621-Artificial Intelligence Course Seminar
Displacement of a Robot
• Currently, Research on “Agricultural robots” is active in Japan and
Korea
V S K Murthy and Singre Pavan
27
CS621-Artificial Intelligence Course Seminar
Conclusion
• Need for AI focus on Agriculture sector is discussed
• Bio-system Derived Algorithms (BDAs) are explored
• Identified intelligent approaches which are useful for
mechanizing complex agricultural systems
• Growing Research and technology should contribute to the
basic amenities in agriculture
V S K Murthy and Singre Pavan
28
CS621-Artificial Intelligence Course Seminar
References:
[1] D.E. Goldberg, Genetic Algorithms in Search, Optimization and Machine Learning. Reading, MA: Addison Wesley, 1989.
[2] J.H. Holland, “Genetic algorithms,” Sci. Amer., pp. 44-50, July 1992.
[3] J.B. Bowyer and R.C. Leegood, “Photosynthesis,” in Plant Biochemistry, P.M. Dey and J.B. Harborne, Eds. San Diego, CA:
Academic, 1997, pp. 49-110.
[4] N. Kawamura, K. Namikawa, T. Fujiura, and M. Ura, “Study on agricultural robot,” J. Jpn. Soc. Agricultural Mach., vol. 46,
no. 3, pp. 353-358, 1984.
[5] Y. Hashimoto and K. Hatou, “Knowledge based computer integrated plant factory,” inProc. 4th Int. Cong. Computer
Technology in Agriculture, 1992, pp. 9-12.
[6] Y. Hashimoto, “Applications of artificial neural networks and genetic algorithms to agricultural systems,” Comput. Electron.
Agriculture, vol. 18, no. 2,3, pp. 71-72, 1997.
[7] Yasushi Hashimoto, Haruhiko murase, “Intelligent systems for agriculture in japan”. IEEE Control systems Magazine, Oct
2001.
V S K Murthy and Singre Pavan
29
CS621-Artificial Intelligence Course Seminar
Thank You !
Questions??
V S K Murthy and Singre Pavan
30
CS621-Artificial Intelligence Course Seminar
Photo respiration
Photosynthesis pathways of
Benson-calvin cycle
V S K Murthy and Singre Pavan
31
CS621-Artificial Intelligence Course Seminar
V S K Murthy and Singre Pavan