Predictive Maintenance Mobile App for Civil Engineering Machinery

Predictive app development for civil engineering

A mobile app based on object recognition and deep learning technologies to automate inspections of moving parts on civil engineering machinery. The proposed approach for solving a business problem and the app itself were developed during a 3-day Hackathon organized by Volvo Group Connected Solutions and Volvo Construction Equipment.


Sigma Software team took part in VCE HACK SPRINT 2018, a hackathon organized by Volvo Group Connected Solutions and Volvo Construction Equipment.

Business Problem to be Solved

The hackathon was organized to generate ideas for solving a nasty business problem: machinery downtime due to lengthy manual inspections of moving parts and replacement in case of their wear-out. The hackathon organizers wanted to get a mobile app that could be used by a person without any technical background to automatically measure certain parts of an excavator undercarriage, such as steel shoe and rolling bodies, to a high precision at the construction site. The measurements are then used to evaluate the wear of the parts and predict the working hours left for the part before it needs replacement.

The accuracy of measurements required to make predictions of the part lifespan is within 1-5 mm. Existing Google and Apple measurements tools have insufficient accuracy: measuring error of 10 mm constitutes the difference between a new and worn-out part. So, there was a need to find other solutions, and over 80 developers, designers, and data scientists from around the globe were selected from more than 200 applicants to come up with their ideas and implementations.

The Team

Sigma Software team of five specialists, including two Machine Learning engineers, a Unity 3D software developer, a designer, and a project manager, decided to select a Deep Learning approach to the proposed business problem and managed to implement it in a working mobile app just in 72 hours allocated for this task.

The Approach

To make precise measurements, the team decided to use an object with known size as a reference. Since non-wearing portions of the excavator part are always in the picture and their size is known, our engineers used the height of excavator track shoe plate as a reference value.

Civil engineering machinery service software wireframes

Our Machine Learning engineers created a Deep Learning model to predict whether the part needs to be replaced immediately or predict the remaining service hours at current wear conditions.

Image Recognition and Other App Features

The model for image recognition was trained on more than 400 images to ensure correct object recognition and accurate measurements.

Object Recognition
Machine Learning - Measurement

The achieved recognition results ensure the required result accuracy and allow diagnostics even in poor lightening conditions.

The team also gave some thought to accompanying features that would make the app handy for users.

Additional features that the app can provide:

  • Authorization for operators
  • Recent diagnostics info
  • Linking of several excavators to a user profile by Product Identification Number
  • Dealers map
  • Settings management

The Design

Our designer created wireframes for all screens and thought through the usability of the app. The app also got an appealing and clear appearance and feel, which contributed to the sense of readiness of the presented solution.

Civil engineering machinery service software wireframes
Predictive Maintenance Mobile App - Design

See more details about designing the mobile app on Behance.

The First Place

During the presentation of the results in the hackathon, the team conducted a live demo of the app and demonstrated it at the best. The jury was impressed by a perfect balance of user approach and technical elements, business perspective of all implemented and suggested features.

Sigma Software team won the first place at the hackathon. See the video shared by the hackathon organizers.


In the short time provided, the team developed a working mobile app for predictive maintenance of civil engineering vehicles and won the first place at VOLVO HACK SPRINT 2018 organized by Volvo Group Connected Solutions and Volvo Construction Equipment.