Enable javascript in your browser for better experience. Need to know to enable it? Go here.

Preventing failures with predictive maintenance based on Machine Learning models

Aerialoop is an Ecuadorian start-up that offers delivery services through drones in the city of Quito, creating an aerial bridge to the outskirts of the city and transforming a traditional delivery route from 35-40 minutes to approximately 5-8 minutes.




Although drone delivery is significantly faster, it does come with some novel challenges to ensure the drones navigate the city autonomously without malfunctioning, due to adverse weather conditions, user errors or mechanical failure. 


With the assumption that drone components could fail during flight and the projected increase in drones and routes, it would no longer  be manageable for their technical expert to manually search the logs for possible sensor anomalies.


The company uses the WingQuad3 drone on the delivery routes, which generates several measurements per second of approximately 800 entries each. A 10-minute flight produces 10,000 or more data points for analysis. With this background, we sought to automate the data processing in a machine learning model to learn healthy and unhealthy flight patterns.




The work with Thoughtworks was carried out to increase flight safety through the use of machine learning.  With the help of Google Cloud Platform, we were able  to execute the automation of log processing, train the machine learning model for anomaly detection, and create a dashboard in the drone Health Management System to visualize if a flight presented problems. With this, a technician can make the decision whether a drone can fly or not in the next flight.

This project was completed through the prototype.lab(), an internal program from Thoughtworks Ecuador, created to help entrepreneurs and artists develop a prototype of their product or validated idea that requires an innovative technological solution. A leading group in the technological field, these services helped Aerialoop to size their conceptual project -  to develop much more concise and structured ideas - and guide them on issues in software innovation.

Working with Thoughtworks was a good experience, as I felt understood by the group when talking about software.
Diego Morales
Aerialoop Systems Manager

The Thoughtworks team managed to concretize the project in terms of advances, challenges and information - reaching the idea that maintenance based on machine learning predictive modeling, is a fundamental piece for the safe travel of the drone. However, to continue growing, it must continue to innovate and develop new algorithms.

I would highly recommend the prototype.lab() project, since technology is going to dominate or is already dominating the world, so all companies must have a technological component to help them move forward.
José Barsallo
Aerialoop Operations Director

How can we help you achieve faster growth?