Therloc / Thermomer
Two projects in one? (Yes and No)
Therloc and Thermomer are two big modules from a complete system whose mission is to increase productivity and performance in production lines.
Therloc is a guidance sytem based in AR (Augmented Reality), it's an assitant to help operators in industries to set up the monitoring and analysis cameras position in an automated way.
The point of reference used to position the cameras is a QR code marker thats contains some information encoded. This QR code have to be pasted in the desired configuration point. The information content break down following:
- Positioning information: The distance from the viewpoint (working point of the camera) and the relative rotation in the three-dimensional space.
- Area of interest information: Position coordinates and size indicating area of interest, This is made in order to optimize information storage besides capturing all screen.There can be one or several areas of interest.
- Concrete information: Adapted configuration information related to several aspects like camera model, ambient conditions and some other variables.
A special marker is needed for first-time configuration.
Thermomer is an AI (Artificial Intelligence) system designed to detect anomalies in the production line. Furthermore, it provides the study and analysis capacity in order to reduce the failure and anomalies presence.
Among both modules there is an underlying layer in charge of connecting them. This is the data layer, a centralized or descentralized (still to be decided) storage system that uses a blockchain network to store the configuration records coming from Therloc, and stores the results from Thermomer.
I have to say this project jumped out from (Santiago Frías) idea of project.
The initial idea was proposed by Santiago, and I was invited to join the team and collaborated defining and summarizing the idea concept.
Eventhought we had serveral long sessions meetings both in person and through videoconference, and I thought the main concepts and development lines were very clear and agreed, at some point suddently he was claiming that I am not aligned with his project idea, and also he don't want to make an hybrid app at all.
Before start the real project coding and development I had to leave the team and that is why I decided to keep going and complete the project by my side. So everything I've done in this project is authentically made by me and build from scratch.
As a result, If You take a look at his project, You can realize the idea is pretty similar.
Inspiration
In production lines equipped with thermal cameras or localized areas in buildings or industries that requires monitoring or thermal analysis, very often need to be configured and that is a tedious and time consumption task.
What it does
In order to help to automate the configuration task, the core application provides an AR based guidance visual interface, to help the operator to position the camera, and a blockchain based cloud storage system for the configuration that ensures privacity.
Furthermore, the system provides a customized AI model to detect anomalies in the prodcution line.
How I built it
The main development points are (they don't need to be solved in that exact order, and can be developed independently using simulated input):
Test data acquisition from an object in the 3D space, position and orientation.
Build the QR recognition model.
Test and experiment with the FLIR SDK.
Develop the Cordova plugin from FLIR SDK.
Application mockup. Describing screenshots.
Thermal imagery dataset generation.
Build the anomalies destection ML model.
Reserach about the most suitable blockchain network and implement commnunication and storage using it.
Build the hybrid application.
Challenges I ran into
One of the main problems to solve was that FLIR does not provide an SDK version for hybrid applications, so because of I decided to build and hybrid app I had to climb that mountain that increases development time.
Accomplishments that I'm proud of
Every step is becoming real as planned! it's a satistaction to develop useful tools for industries in this case.
What I learned
What's next for therloc / thermomer
- Realtime information extraction and analysis for mixed reality video content enrichment.
- Voice orders assistant controlling.
- Fixed QR codes limit configuration to static places, so beacons based indoor spatial positioning could be a great improvement.
Log in or sign up for Devpost to join the conversation.