Bayer Innovation Lab Provides Smart Pest Monitoring Solution, Reduces Architecture Costs by 94% on AWS

Part of the Crop Science division of Bayer AG (Bayer), the Crop Protection Innovation Lab focuses on crop protection as a service. A crucial part of this service is to alert farmers of threats to their fields. Farmers need to monitor their fields daily for pests—but constantly going into the fields for monitoring isn’t always feasible given the number of tasks requiring their attention during peak season.

To help farmers monitor pests in their fields, the Crop Protection Innovation Lab developed the Digital Yellow Trap, a low-cost Internet of Things (IoT) pest-monitoring device. Using image recognition powered by Amazon Web Services (AWS), the device alerts farmers if a pest migration is happening in their fields, enabling them to take early action to protect their crops. By using a scalable serverless architecture on AWS, the Crop Protection Innovation Lab reduced the costs for operation by 94 percent and can offer such devices at a low cost to farmers.

Founded in 2018, Bayer’s Crop Protection Innovation Lab uses customer-centricity and cutting-edge technology to create outcome-based solutions for farmers to deliver a “healthy field.” Because Bayer has a history of using AWS, the Crop Protection Innovation Lab was confident in its choice to use a serverless architecture on AWS from the beginning. “AWS is constantly looking for new digital trends and improving its services,” says Dr. Alexander Roth, head of engineering at Bayer’s Crop Protection Innovation Lab. “That’s why we think AWS is definitely right for us.”

Part of keeping a field healthy is to prevent pest damage to the crop—a complex decision that requires consideration of ecological and economic aspects. To prevent damage to their crops, farmers need to know whether pests have migrated to their fields, and if so, they need to know how many. When a critical number of newly migrated pests is reached within a certain time frame, a crop protection product should be used. To better understand the challenges that farmers face, the Crop Protection Innovation Lab team visited customers on their farms and performed field assessments. “We really wanted to understand the answers to two questions: What are the challenges here? And how can we find a solution to improve and automate the process?” says Matthias Tempel, head of product at Bayer’s Crop Protection Innovation Lab.

Bayer’s solution to farmers’ pest problems came in the form of the Digital Yellow Trap, a low-cost IoT device that monitors fields using image recognition to alert farmers of a pest migration. To develop the trap, the Crop Protection Innovation Lab team employed a variety of AWS services. For training and partially hosting artificial intelligence (AI) models, the team uses Amazon SageMaker, a fully managed service that enables data scientists and developers to build, train, and deploy machine learning models at any scale. “We’ve been using Amazon SageMaker for quite some time, and it’s become one of our crucial services for AI development,” says Roth. “AWS is constantly improving its services, so we always get new updates.” To quantify and qualify threats to farmers’ fields, the team uses AWS IoT Device Management, which makes it simple to securely register, organize, monitor, and remotely manage IoT devices at scale.

Through AWS services, the Digital Yellow Trap can offer new insights to help farmers make better decisions. The device takes multiple pictures per day and uses AI to identify pests—providing results on a mobile app on farmers’ smartphones within minutes. The app then offers farmers guidance on the necessary crop protection actions that they should take. “What sets us apart from analog traps are the community and network effects,” Tempel says. “For example, the devices share their current statuses with neighboring farms as an early warning system.” The feedback from farmers has been very positive. “Some farmers were not willing to give us back the prototypes,” says Tempel. “They’re happy that they don’t need to go to the field for monitoring every day.”

Because farming is a seasonal business, scalability is crucial. The Crop Protection Innovation Lab maintains a lean operation with scalable architecture using AWS Lambda, a serverless compute service that lets developers run code without provisioning or managing servers. “Going serverless is possible instantly on AWS, so we had no trouble there,” says Roth. “And the potential that we saw, especially when scaling or reducing costs, was enormous.” Now, the Crop Protection Innovation Lab’s infrastructure can handle tens of thousands of requests per second. And the lab was able to reduce the running costs of its overall architecture by 94 percent by using serverless AWS services. And by using AWS X-Ray, which helps developers analyze and debug applications in development and production, Bayer can optimize existing running costs and forecast costs for the year. Using the tracing capabilities of AWS X-Ray, Bayer can analyze data workflows in a complex microservices architecture. “With a minimum effort and a few lines of code, we were able to solve the major challenge of providing cost transparency and traceability. This gives us a solid foundation for future infrastructure optimization,” Roth says.

The next step for a pest-free field offering is to predict pest movement in advance. “We need to collect a large amount of data to train additional AI models that predict pest migration. And we can only collect such amounts of data if we have thousands of these devices in the field,” Roth says.

The company plans to distribute 10,000 Digital Yellow Traps to farmers in 2022. Initial testing of the device was limited to one type of crop, but due to demand from farmers, the company is planning to expand its distribution and supply devices for different crops and pests. “The most important part for us is the ability to come up with new ideas that we can quickly prototype on AWS,” says Tempel. “Our success is definitely based on how happy the farmers are.”