Problem Statement/Purpose:

Currently, many farmers plant crops that have grown well on their land for generations. However, climate change is starting to cause sizable decreases in crop yields. Meaning, crops native to certain areas which previously flourished are now becoming less compatible to their environment. MIT professor, Arnaud Costinot, found that roughly 66% of crop yield issues can be fixed by relocating farms or crop-switching.

The purpose of Croptimize is to allow farmers to determine the most optimal crop to grow on a certain plot of land and adapt to climate change’s negative effects on yields. It utilizes a mobile robot unit, a kNN machine-learning algorithm, and environmental data to suggest the best crop(s) to grow in a certain region.

Methods:

First, several data points regarding optimal conditions for growing several crops around the world were collected and put into a CSV database. Next, we assembled the robot. We gathered several sensors to test environmental conditions (soil pH, humidity, above-soil temperature, sunlight intensity, and NPK nutrient concentration). We connected these sensors to an Arduino Mega using various breakout boards, which interprets the sensor data it collects and sends it to a Raspberry Pi through a serial connection. The Raspberry Pi uses the collected data to run the ML algorithm and outputs results to a database. We organized the components in a casing and added a chassis for mobility and autonomy.

The robot was deployed at an open field in Grant County Park to collect data, and we used that data to determine the most fit crops for that plot by using a kNN clustering algorithm. Our algorithm cross-compares 200,000+ data points from our plant database with the data from the field; the plants from the database with the strongest correlation to the field’s conditions were outputted as the most suitable crops. Croptimize provided accurate results due to its constant readings throughout the day, on different areas of the field.

Results:

Croptimize outputted a crop that best fit the tested plot of land. In our experiment, it outputted ‘spinach’ as the optimal crop for the following conditions: pH levels ranging from 6.5 to 7.1, above-soil temperature ranging from 56 to 74 degrees Fahrenheit, roughly 8 hours of sunlight (at an average of 890 lux), and humidity levels ranging from 63% to 68%. When we tested different parts of the park on different days, the robot received different input values and outputted other fit crops for the plots of land. They included artichoke, cucumber, and guar bean.

Conclusions/Discussion:

The promising results of Croptimize demonstrate the potential it has in reversing numerous inefficiencies in producing crop yields. By testing locations around the world, it can help farmers figure out what optimal crops they can switch to, and figure out where they can relocate their preexisting crop, such that it will grow better. Through practices of crop-switching or relocating crops, which Croptimize enables, farmers can adapt to the imminent and negative effects of climate change on crop yields.

Pranav Palleti

Affiliation

Conference Tracks

Refine Results

Croptimize: A Novel Robot to End Agricultural Yield Inefficiencies Through Crop Switching