Three Year Anniversary Blog: Fall Armyworm Challenge Part 2, Technical Details

Feed_Armyworm

I want to mark the week of my three-year blog anniversary by giving you an update from a technology innovation challenge by Feed the Future I talked about at the beginning May. Sadly our innovation entry did not make it to the list of finalists. In the spirit of open science, open innovation, and open data and providing solutions to social problems without worrying about who gets the credit, I want to go more into the technical details of our solution today.

The WormAway system will automate pest prediction and data collection for farmers. One of the biggest hurdles to taking action before the fall armyworm pest reaches crops is lack of timely, accurate and granular data access presented in an understandable format. It’s crucial to take advantage of existing mobile and social networks available increasingly to a larger number of citizens and farmers in Sub Saharan Africa. The WormAway system provides an affordable and simple data collection solution for farmers that leverages existing community monitoring practices with artificial intelligence methods, statistics and mobile communication tools. Affordability of mobile devices, low digital literacy and low literacy disproportionately affect women and small farm holders in Sub Saharan Africa. Our solution addresses the traditional information collection and distribution channels typically held by men by automating data collection and reporting.

An automatic moth detection system (similar to what is described by described by Weiguang Ding and Stephano Maini) is placed on a 1km grid of land. The system is mounted on a pole off the ground in close proximity to the crops and automatically monitors the number of armyworm moths using a pheromone lure, adhesive liner, digital camera and radio transmitter. The system is powered by solar-powered rechargeable mobile phone batteries. A sensor is also connected to the pole and measures the soil moisture. The number of moths and soil moisture content is used in a Bayesian statistical forecasting method (similar to what is described by J. Holt) to estimate the pest outbreak risk level. If rainfall is more than 5 mm in one day and there are more than 30 moths detected by the camera, armyworm is predicted with 80% accuracy at the village level. Subscribing farmers receive a pest outbreak alert text message of Low, Medium or High on their mobile devices via a SMS text message. The alert prompts users to take appropriate action but does not direct farmers what specific actions to take.

wormaway-sketch

Now moving on to a bit of a literature review. There is no existing system that automatically predicts pest outbreak likelihood using a solar-powered system that communicates more than one time daily – accurate, farm-level actionable data to users. Adriano Guarnieri uses pheromone traps, battery, mobile phone and digital images system to count the number of moths and predict outbreaks but is missing the communication mechanism. Weiguang Ding uses deep learning methods for counting pests in images inside of field traps. J. Holt predicts moth outbreak risks by number of moths and rainfall. European Patent Application 2892331 uses an optical sensor to count the number of dead moths and relays that information to a remote operator. Chinese patent CN 101938512 uses pest counts from a remote sensor and a monitoring center but did not include weather data in the pest forecasting. U.S. Patent 7,603,621 provides a mobile computer interface with icons for illiterate users but no feedback to users based on their input. Robert Cheke estimates armyworm moth flying speed and path. The European Research Council is developing a mobile tool with an icon-based mobile data visualization interface and citizen science methods to monitor environmental conditions.

Since neither of the co-inventors (me or my husband) is an expert in the field of agricultural pest management and detection, we’d love to hear your feedback. Would the system be effective and marketable in sub-Saharan Africa? In the next blog, I’ll talk a little bit about issues related to data privacy and production considerations for the WormAway automatic pest detection system.

Advertisements

One thought on “Three Year Anniversary Blog: Fall Armyworm Challenge Part 2, Technical Details

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

w

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.