Indoor Positioning Using the OpenHPS Framework

Maxim Van de Wynckel, Beat Signer
Web & Information Systems Engineering Lab
Vrije Universiteit Brussel

What is OpenHPS?

An Open Source Hybrid Positioning System

What is OpenHPS?

An Open Source Hybrid Positioning System

  • Any technology
  • Any algorithm
  • Various use cases
  • Flexibile processing and output
    • Accuracy over battery consumption, reliability, ...
  • Aimed towards
    • Developers
    • Researchers

Process Network Design

Process Network Design ...

Process Network Design ...

Modularity

Modularity ...

Communication
Socket, MQTT, REST API, ...

Data Storage
MongoDB, LocalStorage, RDF, ...

Positioning Algorithms
IMU, fingerprinting, OpenVSLAM, ...

Abstractions
Geospatial, location-based services, geojson, ...

Other
React-Native, NativeScript, Sphero, ...

Data Processing

DataObject

Absolute and Relative Positions

Absolute

  • 2D, 3D, Geographical, ...

Relative

  • Distance, angle, velocity, ...
  • Relative to another object

ReferenceSpace

An absolute position can be relative to a reference space.

DataFrame

DataFrame ...

Pushing Data

DataFrame ...

Pulling Data

DataFrame ...

Pushing Error

SymbolicSpace

An object that semantically defines a space

  • Spatial hierarchy
  • Graph connectivity with other spaces
  • Geocoding
  • GeoJSON compatibility
  • Can be used as a location
  • Can be extended ...

SymbolicSpace ...

ts
const building = new Building("PL9")
.setBounds({
topLeft: new GeographicalPosition(
50.8203,
4.3922),
width: 46.275,
height: 37.27,
rotation: -34.04
});
const floor = new Floor("PL9.3")
.setBuilding(building)
.setFloorNumber(3);
const office = new Room("PL9.3.58")
.setFloor(floor)
.setBounds([
new Absolute2DPosition(4.75, 31.25),
new Absolute2DPosition(8.35, 37.02),
]);

Location-based Service

Location-based Service ...

Location-based Service ...

Demonstration

  • Indoor positioning use case
  • Use existing techniques
  • Validation of flexibility and modularity

Positioning Model

Positioning Model ...

Positioning Model ...

Positioning Model ...

Positioning Model ...

Positioning Model

Online App

Positioning Model

Online App

ts
ModelBuilder.create()
.addShape(GraphBuilder.create()
.from(new IMUSourceNode({
source: new DataObject(phoneUID),
interval: 20,
sensors: [
SensorType.ACCELEROMETER,
SensorType.ORIENTATION
]
}))
.via(new SMAFilterNode(
frame => [frame, "acceleration"],
{ taps: 10 }
))
.via(new GravityProcessingNode({
method: GravityProcessingMethod.ABSOLUTE_ORIENTATION
}))
.via(new PedometerProcessingNode({
minConsecutiveSteps: 1,
stepSize: 0.40
}))
.via(new VelocityProcessingNode())
.to("pedometer-output"))
.build();

Dataset

Dataset ...

Total BLE Beacons: 11
Total detected WLAN access points: 220
Total stable WLAN access points: 199

Training Test
Datapoints 110 30
Total fingerprints 440 120
Duration (per orientation) 20s 20s
Avg. WLAN Scans (per fingerprint) 6 6
Avg. BLE Advertisements (per fingerprint) 16 15

Validation Results

Static Positioning

WLAN fingerprinting BLE fingerprinting BLE multilateration Fusion
failed points 0 6 12 0
average error 1.23 m 3.23 m 4.92 m 1.37 m
minimum error 0.01 m 0.17 m 0.74 m 0.01 m
maximum error 4.77 m 15.39 m 19.26 m 9.75 m
hit rate 95.82 % 80.83 % 52.50 % 96.67 %

Validation Results ...

Trajectories

Validation Results ...

Trajectories

WLAN + BLE WLAN + BLE + IMU
average error 3.28 m 1.26 m
maximum error 9.60 m 3.10 m
average update frequency 3.04 s 0.52 s

Contributions and Conclusions

  • OpenHPS: open source framework for hybrid positioning
    • Aimed towards developers and researchers
  • Abstractions such as location-based services and spaces
  • Validation of an indoor positioning use case
  • Configurable and interchangeable nodes and services
  • Public dataset with multiple orientations


Visit https://openhps.org for additional resources, documentation, source code and more!