Please have a look:

https://www.ijert.org/application-of-iot-and-machine-learning-in-agriculture

https://arxiv.org/ftp/arxiv/papers/1906/1906.03106.pdf

Use soil, Rain, temperature sensors to collect data and make a dataset as large as is enough Then apply machine learning or deep learning directly to the dataset Note: The testing data will be directly from the sensors (Real-time )

Hi Abburi VM Subbarao ,

Two years back some of my students worked on an identical projects. Find the link for the associated Research papers

Conference Paper User Controlled Precision Irrigation System

Conference Paper Performance comparison of TSA methods using past weather dat...

Conference Paper Crop management of agricultural products using time series analysis

Terms such as "artificial intelligence" and "machine learning" are mistaken, in my personal opinion. Computers are nothing more than boxes filled with on/off switches. They are not intelligent and they do not learn. Rather, they use data identified, collected, and entered by humans. That data is processed by heuristics and algorithms developed by humans. The algorithms and heuristics are composed of equations and parameters that are adjusted according to processes developed by humans. Those processes depend on the data at hand. The outputs of that processing are selected and interpreted by humans.

During an exploration of how computer engineering can have a positive impact on precision irrigation (https://github.com/SoothingMist/Embeddable-Software-for-Irrigation-Control), I used C5 (https://www.rulequest.com/download.html) to build a next-day rain classification model. The result of using that model within a trial with ETc calculations resulted in 20% less estimated irrigation. A paper on that part of my exploration has been accepted for publication.

While looking at a paper by Dr. Geetali Saha, I've become intrigued by the idea of combining ETc with soil-moisture sensors and weather data. Am presently gathering components to try something like that in an outdoor setting.

Peter Raeth , yes, we are still working on something like this to materialize.

In the meantime, Abburi VM Subbarao I have worked upon automation and control of a Hydroponics live project and remotely monitored various observation parameters of it using IoT alternatives. My sessions are available online.

Dear Abburi VM Subbarao

This is an interesting application. We will consider this application in our future directions.

By using Artificial Intelligence, we can solve problems more easily and more effectively. That being said, imagine a scenario where we want to create a smart park. We can use IoT,

1. To Design A Smart Irrigation System

2. To Design A Smart Fertilizer Spreading System

3. In Smart Controlling of Fountains

4. In Quality Check of the Located River in the Park(Not Plausible in All Situations)

5. In Air, Light, and Field Noise Monitoring

6. In Crowd Control

7. In Water Consumption Monitoring

, etc.

Now, this is where the AI comes in. We can control all of these intelligently using an AI. AI can record the data and use it to train itself and perform better. Moreover, these data can be used by research facilities to train AI's or be used in urban planning. Even this so-called data, if recorded properly over the years, can be a very good source to understand environmental changes caused by various parameters, thus helping urbanists to design better plans according to the predictions derived using the recorded data. This data can even be used to predict what species of birds(or other animals in some cases) are most likely to migrate to or from the park. And These are only a very little portion of the work that can be done, All by the help of Artificial Intelligence.

Geetali Saha : Have gathered the parts I need. Working out how to send SMS text messages for IoT applications for the general case. Seems that cellular networks are a lot more restrictive as to what devices are allowed. At the present time, am able to connect to a network and send messages to my cell phone. But, the messages never arrive and an "unknown response" is delivered to the cellular sending unit. Working out the reasons that is happening.

Hi Peter,

So glad to know that you got a lead on this.

But the topic may go out of the bounds of this question here...

I know about a couple of Open Source Apps and Hardware on which I and my student team have worked successfully. And been able to deliver SMS Text messages also.

It will definitely work. let's discuss it separately...

Perhaps. But, the topic does seem like it is a hammer looking for a nail and a furthering of imaginative marketing. While I do think adaptive automation has a large role to play in various application domains, as does node-represented mathematical and computing architectures, I am not keen on the idea of "intelligent" machines. It is better to advance human intelligence and enable human learning instead of fantasizing about creating non-human intelligence.

Abdelhameed Ibrahim, Peter Raeth, Geetali Saha,Ali Khalili and Javad Sharifi Thank you all for your insights in the topic. Actually I have worked in the field of precision agriculture, Agromet advisory Services etc.. in my service. We have with the help of ICRISAT developed an Application called "Meghdoot" which provides weather and agromet advisories at your finger tips. but advisories required to prepare by a team of experts based on the field situation. So this particular condition varies from one field to other field and one farmer to other farmer and in one crop to the other crop. This kind of complex situation the algorithm need to work on its own to take the stock of the situation in every field, crop and based on farmer it requires to analyze and provide the suggestion to the farmers at right time to act fast. During 2001 to 2004 we have developed an automated irrigation facility by linking the motor to a set of sensors planted in the field. whenever the moisture goes below the rootzone the sensor detects and switches on the motor till the moisture reaches to field capacity and turns of the motor. Further, based on the spatial data from a farmers field collected at 15 cm and 30 cm depth, I have developed the management zones which helped the farmer to understand why the yields are ow at one point of his 20 ha land as suggested by our team, he followed the application of various fertilizers and other material to bring back that low yielding path to at par with the other area. These are very simple models and tech demos. but my question is related to the simplification of these complex procedure in the field by automation of entire procedures based on the field level data and support the farmer in his decision.

It is a very pertinent and relevant discussion at this stage, since many researchers and start up entrepreneurs are looking at the option of using the AI and ML tools and IOT for the applications in all the sectors and lot of scope is there in agricultural sector. We have also attempted as suggested by Abburi VM Subbarao sir for making the irrigation automatic and field level testing seems to be good. However the farmers are looking for the options at low cost , hence these technologies needs to be cost effective and many of the farmers who are practicing drip irrigation goes for timer based irrigation, since that is cost effective.