AWS for Industries

Fleet management solutions for connected farms


The agricultural industry faces a significant challenge supplying a growing population with the healthiest and most ethical foods possible. The world’s population is expected to reach close to 10B people by 2050 and the arable land utilized to feed this population is on the decline. The farming industry has been functioning near peak efficiency for years, but it must transform to meet the expected increase in demand. We can’t meet this demand without the application of the latest technologies where appropriate. The last decade has seen perhaps the greatest transition in farming since the introduction of the combustion engine. There are several areas of technological innovations applied to the field of agriculture and one of these is the utilization of smart vehicles and robotics in cultivation.

New technologies allow smart farm vehicles to plow the land replacing the manually driven tractors of today, moving farm produce from field to storage, and moving other farming assets in modern farms. This creates a network of moving vehicles and robots on the farm that need to be managed and maintained. This blog outlines the tools that AWS offers for management of these connected farm vehicles and robots. The target audience is Ag-Tech companies that provide solutions in this space as well as fleet management operators that need remote monitoring and daignostic solutions. The blog describes four important Industry use cases in the Agriculture industry that define the requirements that are specific for the industry. This is followed by an overview of the solution that includes the scope covered, proposed architecture as well as the AWS services used for this architecture. Lastly, we take the use case of predictive maintenance and describe in detail how it is enabled using this architecture and the various AWS technologies.

Industry Use Cases

A survey was conducted by Bear Flag in May 2021, polling 600 respondents. All respondents worked on farmland sized at 1,000 acres or over. Respondents ranged from farm owners and presidents/CEOs to foremen. 73% confirmed that they’ve been in a situation in the past when they had a hard time sourcing skilled worker. 71% of the respondents said that it is challenging to find skilled tractor operators or workers. 91% of respondents say they are concerned about the ever-growing cost of production.

According to the USDA, net farm income is forecast to decrease by 8.1 percent in 2021. One of the contributing factors to this decline is higher production expenses, predicted to increase by 2.5%, to $353.7 billion, by 2022. Most of this decline reflects higher spending on consumable agricultural inputs like feed, fertilizer, and labor. These growing production costs force farmers to raise the prices of their harvest and affect the food market, causing higher input costs and lower outputs. To counter this problem, the use of sophisticated capital agricultural inputs such as smart tractors is key. These smart tractors have advanced capabilities to capture key parameters in a digital format. An easy-to-use digital interface combined with remote control capabilities results in lower costs per field, higher job quality, and much lower insurance premiums. Smart tractors and other robotic assets are also much more efficient than traditional equipment. They can run 24/7 with minimal supervision and work faster. By increasing operating efficiency, agricultural productivity is boosted, and field costs are optimized. Based on this data, we see the following four use cases as being the most relevant for Agriculture technologies to make an impact.


Geofences help you differentiate between cultivated fields. You can assign each crop field up a uniquely colored geofence. This is useful in tracking crop yields and crop rotation. It can also help track activities performed on these crop fields—if they have been harvested, undergone pest control, or ready for plowing.


Using live-tracking, follow your raw agri-products, while in transit. See if they are reaching a designated warehouse or a distribution center on time. Get alerts when your drivers are running late or if they are on an unauthorized route. Live tracking can also help you get insights into where your agriculture fleets or equipment are used and help you maximize resource utilization.

Reports and Alerts

Reports help you monitor key performance indicators like engine hours, area of plowed fields, distance traveled by tractors in a day, Fuel and more. Use reports to monitor fuel and maintenance expenses and distinguish between your profitable and non-profitable assets, and set up alerts that enable you to react quickly and help recover stolen or misused vehicles and equipment.

Predictive Maintenance

Stay on top of equipment monitoring through timely alerts and practical maintenance programs. Keeping up with equipment and agricultural fleet maintenance will significantly improve their performance and prevent unwanted breakdowns. You’ll get alerted when your equipment is due for maintenance—so you can plan accordingly.

Solution Overview


The proposed solution provides event-driven agriculture vehicle fleet management applications in the cloud that help track vehicle diagnostics and health, predict required maintenance and provide recommendations to the farm fleet owners. It will provide the ability to automate your fleet, control them from anywhere and harness the actionable data acquired from these fleets. There are certain requirements that this solution will need to full-fill which we list here:

  • Automation: Providing a single operator the ability to operate n number of different type of vehicles or eventually replacing the operator with a smart monitoring system that learns and operates the assets in the cloud
  • Scale: System should be able to scale up and down based on the number of assets required
  • Orchestration: Vehicles and assets may need to interact and therefore some level of orchestration is needed for that as well as enabling single operators have all needed control
  • Customization: The solution should have flexibility to be customized for specific farms and to integrate with other third-party solution providers as needed
  • Operational Excellence: In order to manage global fleets, the solution needs to have high degree of operational excellence and minimum downtime and high resiliency
  • Security and compliance: The solution should ensure security of data and compliance with local regularity requirements which may vary by geography and region
  • Optimized for remote operations: Many farms are located in regions where connectivity is poor and access to electricity and internet is choppy at best. The solution needs to overcome this hurdle. It also needs to enable the requirements for vertical farming needs
  • Enable Agriculture specific operational functions: The solution should enable or integrate well with agriculture specific functions performed by Ag-Tech solutions such as Variable Rate Technologies for nutrient and pesticide application in farms


The solution architecture below takes a modular approach and describes the various components and services and how they work together to enable such a fleet management solution. We deploy edge solutions such as AWS IoT Greengrass, Amazon SageMaker Edge Manager, SDKs and gateways on the equipment to collect the data and manage the lifecycle of ML models on edge devices. Transport locally collected data to the cloud using either AWS IoT Greengrass stream manager for high-volume data or over MQTT protocol. Using AWS IoT Analytics, further enrich and analyze device data to build maintenance models within Amazon SageMaker. Deploy those models on the edge using SageMaker inference. Monitor the operational health of the vehicle itself and perform predictive maintenance on the data collected from the vehicle. Enable fleet monitoring, tracking and geofencing capabilities through Amazon Location service and AWS IoT Events. The different metrics, graphs and codes can then be shared with the users or operators in any device form like mobile or web platforms. Cloud monitoring solutions help reduce down time and provide for a smooth operation. You can also update and upgrade these vehicles remotely using over the air updates (OTA).

Deep dive: Predictive Maintenance

In this modern technology-driven world, it is business imperative for companies and farmers in the agriculture industry to keep vehicles and equipment (asset) in optimum operating condition. Historically, a significant percentage of overall operational costs have been co-relational to vehicle & equipment maintenance. In an attempt to keep their agricultural fleets in good condition, most companies and farmers relied on time-based maintenance schedules. While these routines do utilize some degree of statistical analysis, they are static and reactive by design. Predictive maintenance is a mechanism that uses ‘Condition based monitoring’ tools and techniques to monitor the performance of a fleet or a piece of equipment during operation. The ingested data and related information help to predict the future asset failure point, allowing for the asset to be fixed or replaced just before it fails.

The use of predictive maintenance in the Industrial sector is well understood and is a use case that has been implemented successfully. The agriculture sector however has been lagging in this front due to some of the differences as compared to the industrial sector. In heavy industries the assets or machines are primarily stationary and within the confines of a factory or shop floor. The assets in the Agriculture sector on the contrary are mostly moving and mostly in rough terrains and sometimes remote locations that have lack of connectivity. The other important difference is the weather and geo factors also effect the Ag sector more than the industrial use case. Till now the assets were not smart assets and were not generating data points as well. If the technical challenges are addressed, the Agriculture sector will see similar benefits from predictive maintenance as the industrial sector. The remote monitoring and diagnostics of farm equipment allows for overall cost-effective operations and saves time on equipment breakdowns. There are remote farms where a technician visit is costly both in terms of time and money. Such farms will benefit the most from the successful implementation of this use case. Some predictive maintenance benefits for different stakeholders include:

  • The fleet vehicles owner will benefit from avoiding lost revenue from long downtimes due to unforeseen repairs and extended asset lifespan
  • The farmer benefits from a more reliable and productive farming equipment by avoiding unexpected downtimes
  • The vehicle manufacturer gains the operating information, and the variable environmental conditions the equipment is exposed to.

Below are some basic steps to implement a predictive maintenance solution.

1. Connectivity – Install IoT Devices

Affix relevant sensor(s) to the asset and equip with required software & hardware component to establish connectivity to a network (AWS Cloud). To meet the need for smart farm assets, LoRaWAN and NB-IoT technologies enable hardware IoT applications to transmit the payload of data to a target platform (AWS IoT Core) over a larger operating range. The LoRaWAN standard ensures that devices are connected even when they are on the move. Furthermore, it allows data to be sent asynchronously which makes it efficient for sending Tele-metrics data to AWS Cloud for storage. Using AWS IoT Core, LoRaWAN devices and gateways establish private connectivity to the AWS Cloud without developing or operating a LoRaWAN Network Server (LNS). NB-IoT does not require a gateway to provide connectivity, instead NB-IoT devices maintain a synchronous connection to the cellular network regardless if there is data present to send. Thus, making NB-IoT more reliable for real-time analytics application

2. Baselining

A key requisite to practice predictive maintenance is establishing asset baselines. We need to monitor an asset’s conditional baseline to set an acceptable thresholds and relevant key performing indicator (KPI) for an asset and collect data with installed IoT devices.

3. Event detection and response – IoT events

Smart vehicles are equipped with multiple sensors for monitoring environmental conditions, hazards, and detecting changes in operations. Use data collected from sensors to monitor the vehicle states and the events that affect these states. AWS IoT Events is a fully managed service that makes it easy to detect and respond to events from IoT sensors and applications by creating a detector model. Leverage data from vehicle sensors like oil pressure, temperature, vibration, and use IoT Events to monitor vehicles using an IoT Events detector model. AWS IoT Events infer the vehicle states based on all the sensor data coming in and take actions (alert the end user) or trigger another service. AWS IoT events integrate with applicable AWS services to events such as:

  • AWS IoT Events to monitor incoming IoT events (GPS coordinates, a heartbeat representing the vehicle is active/inactive) to detect changes in operations
  • AWS IoT Greengrass stream manager to transfer sensor data and images to Amazon Simple Storage Service (S3)
  • AWS Lambda function to request SageMaker to detect obstacle and send a notification message to SNS if obstacle detected
  • Amazon SageMaker to detect physical obstacle in the image uploaded by validating it against a model defined and deployed in SageMaker
  • Amazon Simple Notification Service (SNS) to send notifications to end user
  • AWS IoT Core for vehicles’ communication with AWS

4. Data Ingestion and Intelligence – Store and Learn from Data

Tele-metrics data from devices is stored (Amazon S3), processed & aggregated (Amazon EMR or Glue) based on predictive analytics requirement and train the systems to learn from data (Amazon Sagemaker) to automate analytical model building, identify patterns and make decisions with minimal human intervention, hence developing a customized maintenance solution.

5. Schedule Maintenance – Visualization and Alerting

Customized maintenance solution evaluates real time data payload against established data model patterns. The real time decisioning- reflects the data and respective actions in dynamic dashboards (Amazon Quicksight) to visualize and monitor remotely, and trigger alerts (Amazon SNS) for maintenance teams to assess these anomalies or even integrate with Enterprise Asset Management systems to automatically generate work orders.


Innovation at this scale and speed is leading to the next generation of solutions. A fully automated and intelligent fleet management solution customized for large scale farming as well as vertical farming is key for digital transformation of agriculture. The modular solution architecture described in this blog is highly scalable, and easily deployed to help organizations get started. Although the initial solution focuses on vehicles and moving assets in the farm, enhancing it to monitor and control agriculture specific functions and operations could be the next step. This solution could be readily adopted by different farming practices and scaled as needed.

If you are interested in learning more about Ag-Tech at AWS then please check out our twitch stream content on agriculture the “All in the Field” live stream series. If you would like to get started with building your own solution based on the proposed architecture, please contact us at

Kevin Huang

Kevin Huang

Kevin is a Senior Solutions Architect at AWS working with digital native customers on their cloud native journey. He is passionate about leveraging the cloud to help customers lower development costs and decrease time to market. Kevin lent his expertise predominately in the solution architecture field and has experience in designing and implementing solutions with containers, data analytics, and serverless for small and enterprise customers.

Gaurav Malhotra

Gaurav Malhotra

Gaurav is a Senior Solutions Architect at AWS, working with Startup customers to help them design and architect scalable & secure applications on cloud. He is an Agriculture domain expert and help Agritech Startups to increase business agility through innovation using AWS Services in IoT, AIML, Containers and Analytics. He is also Containers and Kubernetes enthusiast and has worked extensively in Application Modernization, DevOps and large scale Cloud Migrations projects.

Prashant Tyagi

Prashant Tyagi

Prashant joined AWS in September 2020, where he now manages the solutions architecture team focused on enabling digital native businesses. Prashant worked previously at ThermoFisher Scientific, and GE Digital where he held roles as Sr. Director for their Digital Transformation initiatives. Prashant has enabled digital transformation for customers in the Life Sciences and other industry verticals. He has experience in IoT, Data Lakes and AI/ML technical domains. He lives in the bay area in California.