Now we are wrapping up the Global Indoor Navigation (GIN) project and the year 2021. We had good progress both in the project and as a company.
GIN has aimed to develop a new self-learning indoor navigation that automatically characterizes buildings through the use of the service in ordinary mobile phones. New algorithms and methods have been developed that use collected position data and automatically identify the most common routes, stairs / elevators and entrances in a building. With this characterization you can then create a lot of valuable services e.g. indoor navigation, locating people in need, finding equipment and efficiency improvement solutions.
Evaluation of different types of buildings shows that we can automatically estimate paths with approximately 2-10m accuracy, which is fully sufficient for indoor navigation in e.g. shopping centers.
The GIN project has succeeded to develop a complete prototype for self-learning indoor navigation:
A new Android app based on a new Indoor SDK for easy data collection and building survey:This Indoor Survey App makes it possible to create an accurate indoor positioning for a building in just an hour.
New positioning method with deep neural network In the tests at the different sites: office, residential, university and shopping malls, we got about 2-5m median error, a 2-10x improvement compared to our normal trilateration method! This method is planned to be put in production for Combain Enterprise Indoor Positioning product. Thus will provide by far best in class indoor positioning accuracy for surveyed buildings. Accuracy of 2-5m without any extra infrastructure!
New route extraction method that automatically calculates entrances and the most common paths in a building
The accuracy of the paths are about same as the accuracy of the positioning. Thus 2-10m depending on positioning method and building.
New portal that shows building features, calculated paths in a building and demonstrates indoor navigation In the figure we see indoor navigation between 2 random points based on automatically generated indoor routes. We used the open source GraphHopper Routing Engine and created an API that any front end, system or app can use for indoor navigation. Thereby we can enable “Global Indoor Navigation” for almost any building in the world!
Looking back, GIN project was great to have and made us focus on R&D during the pandemic 2020 and 2021 when we could not travel and there were not as much international work to do as we are normally used to. It was all made possible thanks to the financial support of Vinnova, Sweden’s innovation agency. Also our research partner Centre of Mathematics at Lund University was invaluable to reach the research results. We will continue this cooperation even if the resources will be less and completely self financed from now on.
Still 2021 will be our best year ever, both in revenue and profit, despite the tough times and no travel at all. Large new customers and Combain is growing again as we have been used to (we have grown every year since start). 2022 looks very promising and our plan is to continue the research and develop products, built on the results from GIN. We will be shipping first indoor navigation solution as part of the Traxmate IoT tracking platform before 2022(!) to a very large customer. Next year, a new release of the indoor navigation solution will include research results from GIN!
If you are interested to get a next generation indoor positioning and indoor navigation solution, do not hesitate to contact Combain!