Predicting calmness in smart spaces

YIT
  • Sensors
  • Predictive models
  • IoT
  • Time series forecasting
Problem

Problem

Current smart buildings cater for efficiency, but the users want to find spaces where they can be comfortable.

Solution

Solution

CalmCast calculates and predicts space calmness and activity, helping users to find the ideal space suitable for their needs.

How does it work

Results

  • Users find the space they needSmart space users find the spaces they need with less effort and have the possibility to even relax at the office
  • Spaces can be optimized to fit user needsWhen calmness is measured, spaces can be optimized to fit their usage
  • Spaces can adapt to changesCalmness prediction can help compensate for large crowds with reactive air conditioning, lighting or soundscapes
  • Highly integratableThe Calmness prediction can be integrated with already existing applications which makes user onboarding easy

Finding the calm with sensors and machine learning

Most smart buildings, especially offices, have been optimized for efficiency and productivity. Open space layouts, hot seats, and easy room booking make for effortless and swift logistics, but leave the responsibility of finding an appropriate work space to the office user. Have you ever hurried down the office hallways, looking for a quiet room to take that urgent and sensitive phone call? What if you knew exactly where to find the space, without having to look for it by going from door to door? With CalmCast, future smart office spaces could help the users increase their productivity and comfort by showing when different rooms are calm enough for relaxation or focus, and when they suit socializing and networking needs.

In order to broadcast and forecast the calmness of spaces, the activity must be measured. Calmness Index (CI) is a metric assessing how calm or active a space will be during a given time interval: for example right now, or an hour into the future. The calmness measurement can be based on a variety of smart sensors such as human presence, sound levels, space temperature or light usage. The continuous sensor measurement makes a great basis for accurate calmness and activity forecasting.

The CalmCast solution was tested in a KEKO Proof of Concept project, between fall of 2020 and spring of 2021. During the PoC period, a Helsinki based office campus Maria01 was used as the CalmCast test environment, as it provided an excellent platform for the sensors vital for computing the Calmness Index. Within this test environment CalmCast was able to run in real time, and learn to predict the office’s calmness. KEKO service provider Proximi.io took the calmness data into use in their KEKO Wayfinder application, which is available now in the Play store.

In the future, KEKO smart buildings can utilize the calmness data and adapt the building accordingly. For example, the building can increase ventilation in rooms that are exceptionally active and make them more tolerable to users. Additionally, the space calmness data could be made available to other service providers from outside of KEKO, such as space reservation systems & wellness applications. These external service providers could then utilize the Calmness Index when recommending suitable working spaces to office users in different situations.

Summary

What we did
  • Service design
  • Software development
  • Machine learning
  • Cloud computing
  • MLOps
  • UX/UI
Data science
  • Time series forecasting
  • Latent variable discovery
  • Tensorflow
  • Keras
Language & Tooling
  • Python
  • Docker

Ask for more

Vili Hätönen

Vice CEOvili@emblica.com
Vili Hätönen

Juhana Laurinharju

Software Developerjuhana@emblica.com
Juhana Laurinharju
Do you have a probem you would like us to create a solution for? Contact us and lets talk more!

See more projects