Problem

Whenever we have visited a farm, we have noticed that a lot of farming activities are labour-intensive. We, ourselves, started growing spinach and radish on a farm to validate the problem statement and the solution. We noticed that weeding and irrigation are two activities that need to be conducted all year long. There are other processes in the farm that are very important to the future of the crop, like the incubation of seeds before planting and the effective making of compost. These activities have to be performed with much precision. It has been found that during seed incubation, even a 5°C change in temperature can wreak havoc on seed germination. Also, the huge amount of compost being produced in the farm can release high amounts of greenhouse gasses like methane, ammonia and CO2 into the atmosphere.

Weeds are unsightly plants that decrease harvest efficiencies. Because they are plants competing with crops to survive, weeds are actively removing nutrients from the soil and leaving crops with lower amounts of key nutrients necessary to maximize yields. Currently, farmers employ costly manual de-weeding or spay weedicides over the whole field, which adversely affects the soil and nearby water bodies.

Irrigation of crops at the right time so that they are neither too dry nor waterlogged is also very important. Soil moisture is a  parameter in the global energy and water cycle that is important to irrigation. Its measurement is usually performed by taking the mean of soil moisture levels in different parts of the field. This is not feasible with in-situ measurements (placing many costly moisture sensors around the field), due to financial and physical constraints. So, crops are often irrigated rashly, causing flooding and damage to crops. 

Solution

Agricultural Autonomy mechanisms are heralded as the future of farming and AutoFarmer is one such system. It is an autonomous system of systems that enables a self-governed farm with minimum supervision. It uses Robotic automation, artificial intelligence, and satellite data to help the farmer in decision making and to remotely survey the farm.

Autofarmer promotes sustainable agricultural practices in seed incubation, weed control, irrigation, and compost creation. It consists of many subsystems:

 - The AutoFarmer robot uses machine learning models to monitor your farm. The robot runs on the rows in the farm and takes pictures of the crops using a camera. These pictures are fed to a raspberry pi which is used for surveying the plants for growth progress, disease detection, or harvest readiness. It can also do actions like detecting weeds and spraying weedicide on them.

 - A ground control watering system monitors and irrigates your farm by processing a variety of sensor data from your farm and satellite soil moisture data.

 - A seed incubation system that helps you effectively monitor your seeds for temperature, moisture and light intensity for optimal germination before planting them in the farm

 - A smart compost bin system monitors the maturity of the compost made on the farm based on the gases emitted.

 - A camera system that you can access live footage of the farm from anywhere.

All the data and camera footage collected by the AutoFarmer system along with manual overrides for the equipment is present on a central monitoring dashboard that the farmer can access from anywhere. This prototype dashboard is cloud-based using Thinger.io IoT Platform and shows data from all AutoFarmer systems like the data from various sensors like DTH11 temperature and humidity sensor, soil moisture sensor and LDR for light intensity. 

We can develop an open-source web application for the monitoring dashboard but for this prototype, we have used an IOT Platform. This web application can be deployed on cheap microprocessors like Raspberry pi which the farmer can use to monitor his farm parameters and alert to his mobile when required. 

For the satellite data, we have used agricultural monitoring API and Google Earth engine data to provide visual moisture levels. Based on these micro-climate parameters, the irrigation schedule can be automatically determined for large farms.

The details of the working of the robot and the ground control unit are described in the video. 

We planted spinach and radish plants on a farm in the outskirts of Bangalore. After one month of growth, we tested our robotic solution on the farm.