AiLE— Advanced indoo.rs Localization Engine

Silvia Pichler, 
/wp-content/uploads/2018/06/AiLE-1-e1534982640127.png

Global Navigation Satellite Systems (GNSS) such as GPS have been around long enough to become a natural part of our everyday lives, and in mobile applications we commonly rely on or provide Location Based Services (LBS) such as Google or Apple Maps, Uber or Foursquare. However, in urban and indoor environments the otherwise ubiquitous GNSS are unavailable due to severe Non Line Of Sight (NLOS) conditions — as there are no direct satellite signals because of buildings, walls, and other obstacles.

AiLE - LOS obstruction

Fig. 1: GPS signals blocked by buildings (NLOS)

 

The essence of indoo.rs business is to provide location as an accurate and reliable blue dot on customers indoor maps. Instead of satellite signals, indoor Location Based Services (LBS) must rely on other signals, which  for modern smart phones means on-board radio (WiFi, Bluetooth, cellular) and motion sensors (accelerometer, gyroscope, magnetometer). However, getting an accurate and reliable location from these signals is far from straightforward. The Advanced indoo.rs Localization Engine (AiLE) is our approach to this problem, and it is what sets us apart from the competition in on-board mobile localization [1].

 

There are three main components to AiLE:

    • FLIP – The FLexible Indoor Positioning system that provides absolute positioning based on iBeacon radio maps [2]
    • PDR – The Pedestrian Dead Reckoning engine that estimates relative motion by detecting user steps with length and heading based on motion sensor data readings [3]

 

  • AKF – The Adaptive Kalman Filter (AKF) that fuses FLIP positions and PDR steps to a smooth trajectory [4].

AiLE - LOS obstruction

 

We will give an outline of these components below, and dive into the details in further posts.

FLexible Indoor Positioning estimator

FLIP, or the “FLexible Indoor Positioning estimator”, uses the radio data obtained from the smartphone and processes it through a dynamic radio buffer to compensate for missing signals. This data is then compared to the Gaussian process smoothed radio map in several iterations to pinpoint the location and uncertainty.

AiLE - FLIP map

Fig.2: FLIP Map

 

Pedestrian Dead Reckoning

Pedestrian Dead Reckoning, also known as PDR, is used to remove all redundant mixed-in phone data noise. It then leverages the smartphone sensors to accurately estimate motion and direction through detecting and measuring steps while walking. The derived data output is given in x, y and z coordinates.

 

Whereas the data obtained from the vertical sphere is used for step detection, horizontal data, obtained from the beginning of one step to the next, helps estimate the direction of the step. Finally, the length of a step is then calculated via machine learning system using accelerometer and gyroscope data.

 

Adaptive Kalman Filter

A basic Kalman filter fuses qualified step events with qualified radio locations to produce optimal combined positions. The Adaptive Kalman Filter, AKF, is an extension and the third essential component of AiLE. It fuses the resulting absolute position data from FLIP with the relative motion data from PDR into a trajectory to generate optimal combined positions. Potential non-linearities are compensated, unknown radio location and PDR qualities are estimated, and as a result, smooth and unbiased user position trajectory can be obtained.

 

After merging and processing the data streams in the AKF, the resulting accurate real-time Indoor Positioning Data is then transmitted to the smartphone, appearing on the map in the form of the Blue Dot, helping the user navigate. The Positioning data is also saved to the indoo.rs cloud, where it can be retrieved as valuable analytical data.

 

The radio data is selectively collected from the AiLE terminal and forwarded to the indoo.rs cloud, where it is processed through the indoo.rs SLAM (Simultaneous Localization and Mapping) Engine to create accurate radio maps. These SLAM algorithms take on a critical role in indoor radio mapping: By automating a major part of the procedure, rendering almost all manual input obsolete, the mapping procedure can be sped up by 90%. This helps  save valuable time and money and takes away inconvenient and complex working routines.

 

The result of the interplay of all these components, AiLE, is state of the art Indoor Navigation, Asset Tracking and Analytics.  

 

If you would like to learn more about AiLE, have a look at the below referenced papers for more in-depth reading on the topics.

 

Acronyms

  • AiLE — Advanced indoo.rs Localization Engine
  • AKF — Adaptive Kalman Filter
  • GNSS — Global Navigation Satellite System
  • KF — Kalman Filter
  • LBS — Location Based Services
  • NLOS — Non-Line-Of-Sight
  • PDR — Pedestrian Dead Reckoning
  • SLAM — Simultaneous Localization And Mapping

References

  • [1] Grizzly Analytics, “GEO IoT World indoor location testbed report”, 2016, http://www.grizzlyanalytics.com/report_2016_06_testbed.html
  • [2] R. Müllner, T. Burgess and H. Schmitzberger- FLIP-Flexible Indoor Position Esitmator
  • [3] A. Ettlinger, H. Neuner and T. Burgess – Smartphone Sensor-Based Orientation Determination for Indoor-Navigation, In: Gartner G., Huang H. (eds) Progress in Location-Based Services 2016. Lecture Notes in Geoinformation and Cartography. Springer, Cham
  • [4] T. Burgess – End users used to GPS navigation, expect the same service indoors, article for coordinates magazine, Aug 2016.
  • [5] B. Dong, T. Burgess – Adaptive Kalman Filter for Indoor Navigation
  • [6] A. Ettlinger, H. Neuner and T. Burgess – Orientierungsberechnung mit Smartphone-Sensoren, In: Lienhart W. (ed) Ingenieurvermessung 2017. Beiträge zum 18. Internationalen Ingenieurvermessungskurs Graz, 2017. Wichmann, Berlin

 

You may also be interested in: