The project aims at developing an efficient farm management system that would prove to be resourceful in identifying pests on crops, report the nutrient and moisture content of soil, and accordingly irrigate the farm. For an efficient implementation, the system is integrated in a moving bot whose brain is the DE-10 Nano Cyclone V SoC FPGA board. For the time being, we are implementing our design on kharif crops, which can be further expanded to include rabi crops as well after suitable changes. For an efficient implementation, our project has further sub divisions as controlling the wheels of the bot, path tracking, analyzing the nutrients of the soil, image processing for detecting pests and irrigation system.

Project Proposal

1. High-level project introduction and performance expectation

The project aims at developing an efficient farm management system that would prove to be resourceful in identifying pests, report the nutrient and moisture content of soil, and accordingly irrigate the farm.  For the time being, we are implementing our design on kharif crops, which can be further expanded to include rabi crops after suitable changes. 

The DE-10 Cyclone V FPGA has an USB to UART converter which is extremely useful for interfacing external devices using USB. The Arduino headers enables us to interface with l293d with much more ease. These header pins can be used to further expand the field of our project in the future by adding a pump to spray pesticides in the field as well.

Motor Control: 

We will be using the l293d module interfaced with the FPGA to control the speed of the motors, which will be used to drive the rover. The motor driver takes PWM input to control the speed of the motor. To generate PWM, we would take duty cycle as the input. We will run a mod-100 up counter. Upon reaching the required percentage value of Ton, we will toggle the waveform to 0, thus providing us with a PWM signal of the required duty cycle. 

The bot would need to take turns. We would implement this by wheel locking technique. To implement this, we would need to lock the wheels in the direction which the turn is to be taken; the opposite wheels are to be kept running. For example, to take a left turn we would lock the left wheels and run the right 2 wheels. 

Path following: 

A fixed path would be set up around the field which the rover would need to follow. We would be achieving this by making use of a color sensor, CN0397 module, to detect a specific color and keep the rover lined with the path. If there is any kind of deviation from the path, the bot will stop and turn each side for a specific angle till the path is detected. This angle would be kept at 90 degrees preferably to facilitate turns as well. The path would be of a color which is uncommon in the field. CN0397 has RGB sensors which can detect the intensity of these colors. We will compare these intensities and identify the color based on it. 

Nutrient Analyzer 

An important factor that determines the health and quality of crops is the nutrient content of the soil. A plant should get the required nutrients in the right amount, so our bot is equipped with a NPK sensor whose values are input to the FPGA board. These values are processed and accordingly the user is sent prompts regarding the nutrient content of the soil. 

Image processing unit of bot 

One of the major issues farmers faces is the pest infestation. An image processing unit would help look out for the infected crops. The idea is implemented with the help of vision HDL toolbox that provides pixel streaming algorithms for the implementation of such a system on FPGAs. The advantage of using vision HDL toolbox is that it helps generate readable and synthesizable code in Verilog. The steps involved in image processing are:

1. Convert the frames of the video feed into a stream of pixels.

2. Edge detection, used for image segmentation and data extraction. 

3. Corner detection which gives a better resolution. 

4. Conversion of pixel stream back to frames. 

Irrigation System 

Irrigating farms can become a tedious job for the farmer. The bot is equipped with a moisture sensor that would help decide to irrigate the farm, when the moisture content of soil falls below the required value. The output of the moisture sensor is input to the FPGA board which in turn communicates wirelessly with an Arduino board that controls the servo action of the water channels. For wireless communication the FPGA board uses the integrated wireless module whereas esp8266 is interfaced with the Arduino board. A servo motor is controlled by the Arduino for each row of the farm. Whenever required, water is supplied in each channel by servo action. The track of each row is kept with the help of accelerometer reading.

2. Block Diagram

PWM Generator

Path detection

NPK analyzer

Image processing

Irrigation System

3. Expected sustainability results, projected resource savings

A small farmer in India owns 1.1-hectare land, which they survey frequently. This survey, although effective, can be improved upon by making use of this bot. For a farm size of 100x110 m, a farmer can take anywhere from 25 minutes to an hour for this survey. Our bot can do the same survey within 40 minutes and at the same time, check for pest effectively and irrigate the land. The data from each survey would be stored on board as a log file and can later be retrieved by the farmer. This would mean more free time for the farmer to do other farming related tasks. 

Irrigation in fields is done by releasing water from canals. This method leads to irregular irrigation and causes wastage of water at the same time. By using our bot, this wastage could be reduced as it would check soil moisture at a specific distance and would release water only for the specific track, so that water could be saved and irrigation could be made more effective. Our bot would also check for the pH of soil as well as the nutrients present. Based on this, it would alert the user, enabling them to take the required step. 

Pests are a big problem in agriculture, accounting for up to 40% losses in edible crops. Our bot would check for pests and would alert the farmer if a pest is detected on a certain plant. This would act as an early warning system so that the said plant could be removed at the earliest, while minimizing the damage to other plants nearby.  

Moreover, using a battery-operated device for this surveying operation would be better than some present methods employed where this surveying is conducted by tractors. Huge amount of fuel would be saved by employing a battery-operated bot and it would be cheaper than buying vehicles. 

4. Design Introduction

5. Functional description and implementation

6. Performance metrics, performance to expectation

7. Sustainability results, resource savings achieved

8. Conclusion