5 AI in Agriculture

Transcript

Brief Introduction

Hello friends. In our previous discussion, we talked about expert systems. And I gave you some of the examples of expert systems. Maybe the Rice Doctor or the expert system developed by the Tamil-Nadu Agricultural University. The expert system developed by the Indian Council of Agricultural Research in the name of Maize agridaksha. So I hope by now you must have tried these expert systems, and how the solutions are being obtained by the farming community as well as the students as well as researchers by these expert systems.

In this class we will be discussing about the Artificial Intelligence, and how it can be utilized for the purpose of providing extension services.

Artificial Intelligence

Artificial intelligence is the ability of a computer or computer enabled robotic system to process information and produce outcomes in a manner similar to that to the thought process of humans in learning, decision making and solving the problem. The similar conditions that we are trying to create here in the machine as it is analogous to the brain of a human-being. So it keeps on getting the inputs from the machine and it behaves like the human brain, and it tries to provide you the solutions.

And how it works. So let us look into that. So Mcarthy who is considered the father of artificial intelligence said that, “It is the ability of the machine to perform cognitive function, we associate with human minds such as perceiving, reasoning, learning and problem solving”. So these processes till day are the domains of the human-beings and they are limited to the human brains. Similar systems are being developed with the help of machines. So that’s what is the concept of artificial learning.

AI in Agriculture

So when we look into the need for artificial intelligence, in case of agriculture. The available statistics says that,

• Holds the promise of meeting the demand to produce 50% more food, and cater an additional 2 billion people by 2050 as compared to today. So this is what is the extrapolated data. By 2050 we need to almost double our food production, wherein we need to take the help of artificial  intelligence. Then only we can feed the growing population.
• It has the potential. Artificial Intelligence has the potential to address the challenges such as
  • Inadequate demand prediction. Because it is bit difficult process to estimate the population as well as the demand that are emerging out of this population growth.
  • And lack of assured irrigation, which has become the global problem not only in Indian context. So everywhere water is a scarce resource, and how can we make the best utilization of this water as a resource specially for cultivation of crops. So wherein artificial intelligence is going to help us.
  • Then overuse or misuse of the plant protection chemicals is another important issue wherein we can take the help of the artificial intelligence.
• The improved crop yield maybe
  • through real time advisory that we are going to provide to the farmers, wherein artificial intelligence can play a dominant role.
  • Then advance detection of pest attacks. If you have the prior information, you can take the precautionary measure. So that the plant protection chemicals can be used at its minimum, so that improving the health of the end-users.
  • Then prediction of crop prices is one of the important issue, specially in case of predominantly agricultural nations like India. And to inform about the sowing practices. Because the date of sowing is very crucial. As per the studies, so if we delay the sowing by a particular day, so that is going to end with the losses in case of yield. So that’s what we need to take care of.

Insect – Pest prediction enables farmers to plan

• The artificial intelligence enabled App developed by the Microsoft and United Phosphorous Ltd, which is known as the pest risk prediction App, which predicts the attack of the Jassids, thrips, whitefly and aphids.

So these are the prominent insects, what the farmer is using huge amount of chemicals to control these things. Without spraying chemicals it is very difficult for a farmer to control these insects. In such circumstances if you have the prior information, we can prepone or postpone the date of sowing, so that we can miss this attack of pests. Or there are many other manipulations that we can take up in crop management practices. But there should be valid data as far as the prediction of the occurrence of these insects are considered.

• So it helps to take preventive action, then provide guidance on the probability of the pest attacks. So that is how the farmer can take the decisions.
• So if you look into this artificial intelligence enabled App. So more than 3000 farmers with less than 5 acres of land in 50 villages across Telangana, Maharashtra, Madhya Pradesh are receiving voice calls for their cotton crop. So which is one of the major cash crops in these regions, in the central part of India including Madhya Pradesh, Telangana as well as Maharashtra, which is known as the cotton bowl.

So you can save huge amounts of money. Money is not the only criteria. The health of the farmer, and health of the soil as well as the nature that also we can protect with the help of such predictive mechanisms. And another perception what we have is the small and marginal farmer, it is a bit difficult to use these information-communication technologies.

So that is how just I would like to draw your attention. So more than 3000 farmers with less than 5 acres of land. Means they are all the small farmers that we can say. So who are making use of the best possible use of this artificial intelligence techniques, so that they are predicting the insect attacks.

• The calls indicate the risk of pest attack based on weather conditions and crop stage in addition to the sowing advisories. So these are the additional services that the App is giving. The artificial intelligence systems are giving.
• Then the risk of classification is high, medium and low so that if the risk is high there are different set of precautionary measures that you can take. If the risk is medium and low accordingly, you can take the precautions and ultimately you can save the crop.

Crop Yield Prediction Model

• Then coming to the next model that is the Crop Yield Prediction Model, which is developed by the NITI Aayog in collaboration with IBM. So a govt institution and a private institution, they are coming together to develop an artificial intelligence mediated system for predicting crop yield.
• Basically IBM’s artificial intelligence model for predictive insights include,
  • To improve the crop productivity.
  • To control the use of agricultural inputs. Maybe it is seed, insecticides, fertilizers, so on and so forth.
  • Then early warning on pest and diseases outbreak, then using the data from ISRO. So this is one of the key feature of this crop yield prediction model, that we can say. So satellite data are already available. So but how can we make best use of this data, using this artificial intelligence is the application models that we are coming out with. So with the help of these models, yes we can make excellent use of it.
  • Then we can develop the soil health cards using the artificial intelligence.
  • And using IMD weather prediction and the soil moisture, temperature. So we can schedule our irrigations etc.

So these are some of the uses of this Crop Yield Prediction Model.

• This particular crop yield prediction model is being implemented in States like Assam, Bihar, Jharkhand, Madhya Pradesh, Maharashtra, Rajasthan and Uttar Pradesh.

The northern as well as the eastern part of the country are being benefited by this artificial intelligence model developed by NITI Aayog in collaboration with IBM, that we can say.

AgroPad : Chemical Analysis of Soil and Water

Then coming to another example of artificial intelligence, which is known as AgroPad, which is basically used for analyzing chemical aspects of soil and water.

• It enables real time, on location; chemical analysis of soil and water.
• A drop of water or the soil sample is placed on the AgroPad, as we can see in the photograph.
• So wherein the microfluidic chips, which is inside the card, it performs on the spot chemical analysis of the sample and provides results within 10 seconds
• In the form of development of these circles. So these circles are the outcomes of the result.
• But how to analyze these circles. So there is an App developed for the purpose. You need to take the photograph using your smartphone. And with the help of that App the results are also available to you, so that it gives the results of that chemical analysis based on the circles and their color.

Because the colorimetric techniques are being used there. So using those techniques it is giving us the results. Yes this type of chemicals are prevalent in this sample. And this is whether it is in a high range or medium range or a low range, and accordingly the precautions are to be taken. So this is what is about AgroPad.

AI Sowing App

Then coming to artificial intelligence sowing app.

• So developed by Microsoft in collaboration with ICRISAT. ICRISAT is an institution located at Hyderabad. So both these institutions put together developed this sowing app.
• It sends the sowing advisories to the farmers on the optimal date of sowing. As we have already discussed the importance of the date of sowing. So if we miss that particular recommended date of sowing. So subsequently we are going to lose a huge amount of yield in terms of reducing the yield. That is why the date of sowing is very crucial decision the farmer has to take in performing the crop production.
• So then for this purpose you need not have any sensor to be established in your field. So you are free from that, but you need a phone for receiving the text messages. Because this is so systematic model, only if you have the phone you will get the information.
• The advisories contained the essential information including
  • The optimal date of sowing, so that you can plan accordingly
  • Then soil test based fertilizer application. So once you add your soil testing results. It gives you the dosage that need to be applied as far as the fertilizer is considered.
  • Then the amount of farmyard manure that you need to apply for your field, again that is based on the soil testing results.
  • And the seed treatment and what are the alternatives that you can think of for the purpose of seed treatment. That is also being sent to you.
  • Then the optimum sowing depth, based on the soil moisture, texture; number of factors that it is analyzing, and it is trying to provide you the advisories.
  • So as on date more than 3000 farmers across the States of Andhra Pradesh and Karnataka are getting the benefits of this
  • And as per the studies conducted. The reason increase in yield of up to the extent of 10 to 30% increase in yield, because of the advisories that are being provided by artificial intelligence sowing app.

AI and Challenges in Agriculture

There are certain challenges in application of artificial intelligence in case of agriculture and allied sciences. So which include,

• The lack of familiarity with the machine learning solutions among the farmers. So farmers still need to be educated on this. How to make use of these applications. So that they can harness the benefits of this
• Then exposure among the farming communities to the external factors like weather conditions, soil conditions and presence of pest is quite a lot, so that is why many times the advisories even that are being provided by these artificial intelligence based systems, so they may not work under certain extreme conditions because of the biological nature of the production.
• Then artificial intelligence systems also need a lot of data to train machines and make precise predictions.

So this itself is posing a challenge. But a number of experiments that are going on, and number of successful results are also before us. So that encouraged by positive results, the scientists are working on these models. As we have already said many times that agricultural production process being a biological in nature. It has its own gestation period. So it is not that we can generate all the data in one week or one month or so. So it needs to take at least depending on the crop, 90-100 days and in some cases 120 days. So we need to have that gestation period to develop one set of data. And many sets of such data are essential for the development of this system. So that is how it is taking lot of time to come to the appropriate conclusion.

Conclusion

Then to conclude this topic on artificial intelligence in agriculture

• The future of farming depends largely on adoption of cognitive solutions. So far the solutions what we were getting from the experts or the research system. So now they need to be replaced with this knowledge based solutions. Because the farmers are also having enough information as the scientist is also having. But we need to have some advancements in this sector, then only there will be a sea change in the agricultural scenario.
• In order to explore artificial intelligence in agriculture, application need to be more robust. So more and more number of farmers should come forward and more and more number of institutions should come forward to make this system more robust, as we have already said. It needs a huge amount of data over a period of time. Geo-spatial, so over a large geographical area. As far as over a period of time we need to have huge amount of data. That needs to be added to this. Then only we can come out with the beautiful solutions using the artificial intelligence.

With this we are concluding this discussion on artificial intelligence, specially in case of providing the extension services. And in the next discussion we will be talking about ‘Bring your own device’ which is an innovative way of teaching as well as learning, and making the professional more effective, in the next week.

Thank You.