Technological developments are enabling data capture and automated processing in the geo-domain to take place at unprecedented scale and costs. As one of the world’s leading digital airports Schiphol is looking to increase their level of process automation, for this reason they developed the Pointcloud Innovation Challenge.
Schiphol tasked six companies with the following goals:
- Discover current technological possibilities, developments and challenges
- Identify current technology for a POC/Pilot at Schiphol or at Heijmans/BAM for existing use cases
- Discover new uses cases related to Pointclouds
- Identify leaders/visionary companies to work with in the future
- Have fun and learn as an ecosystem
Challenge Statement
“How can we use technology to combine Pointclouds, photo’s, BIM and/or GIS models delivering new or improved products and services to people working at Schiphol airport?”
Our Approach
We set out to show how our BrainMatter platform could empower Schiphol and its main contractors with on-time performance during the lifecycle of Schiphol’s physical assets. The following value statements were the ones we set out to achieve:
- Augmenting and automating labor intensive stages of the asset monitoring process,
- Ensuring accuracy and consistency across all assets as well as an intelligent digital twin,
- Establishing cross-domain collaboration whereby multiple groups can use the same platform, particularly with main contractors,
- Ensuring ease of exchange of the annotated data with Schiphol, with the goal of improving on time performance.
Intelligent Automation for Asset Monitoring with BrainMatter
Intelligent automation —the combination of artificial intelligence and automation— changes the way Asset Monitoring is done. Intelligent automation systems like BrainMatter enable the detection, localization and recognition of relevant business assets and monitoring of their status. Anomalies and updates in status of assets are typically connected with existing dashboards or ticketing systems for asset managers.
Automated asset monitoring is driven by recent progress in Artificial Intelligence, and particularly deep learning, which has resulted in a variety of sophisticated and effective models being developed and published in the research community.
Notwithstanding this continuous progress two main hurdles limit the widespread adoption of AI models in businesses: the lack of vast amounts of relevant, reliable, auditable and annotated datasets; and the lack of skilled engineers able to develop and maintain AI models.
Our BrainMatter platform addresses both hurdles.
Firstly, BrainMatter’s proprietary AI-powered sorting and Model Assisted Labeling functionality enable the creation of high quality datasets, ready for the creation of machine learning models, at a very rapid pace. In a typical object detection labeling task, BrainMatter enables a 10-50X speed up factor compared to competing labeling interfaces. All data collected and labeled within BrainMatter is auditable, strengthening the proprietary data advantage a business accumulates over time by investing in curating its own data.
Secondly, its user interface and workflows are designed to ensure any employee with relevant domain knowledge for the business challenge at hand is able to contribute by labeling assets. The creation of machine learning models is handled by the platform, therefore diminishing or completely eliminating the need for an expert team of machine learning engineers.
Day 1 – Initial Dataset Annotation
For the Schiphol case, this results in the fact that domain experts for particular assets at Schiphol, along with their contractors, are not required to have any AI knowledge in order to contribute to the creation of powerful AI models. They can focus on enriching the data by labelling each asset with their domain knowledge. The end result is broader participation, faster creation, and rapid improvement cycles for AI model training.
In just 7 days, we developed a solution that continuously monitored maintenance-related assets in Schiphol Terminal 1. During the challenge 10 members of our team used BrainMatter to annotate nearly 2.000 high resolution panoramic images with approximately 25.000 relevant assets, working only a few hours each day. While everyone on our team had a great time, the core of these amazing results comes from the effectiveness of BrainMatter.
Day 7 – Model Review & Fine Tuning Trained Models
We scored the highest number of points with the judges for showcasing immediate value for Schiphol and it’s main contractors, high feasibility and applicability, scalable and intuitive UI, along with a strong commercial approach. Intelligent Automation was demonstrated for eight assets, however this will soon be expanded across all assets at Schiphol. 
Schiphol-asset-monitoring-dashboard-mockup
This level of automation means that asset managers and quality inspectors no longer have to carry out manual inspections and will instead be notified via tickets when assets such as lighting, smoke detectors and surveillance cameras require maintenance. This allows Schiphol and its main contractors to Intelligently automate asset monitoring much more efficiently, improving first time fix rate, increasing information exchange, saving costs and better identifying risks.
As field technicians become more efficient, they can manage more assets with high quality, spend less time on inspections, and reduce operating cost.
If you’re interested to learn more about BrainMatter, reach out to our team to request a demo.
