advanced indoor navigation - ai ukraine · advanced indoor navigation i.m. gorovyi ceo at it-jim,...
TRANSCRIPT
![Page 1: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/1.jpg)
Advanced Indoor Navigation
I.M. GorovyiCEO at It-Jim,
PhD, signal and image processing
![Page 2: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/2.jpg)
OUTLINE
• Introduction to Navigation
• User Location in Buildings
• Indoor navigation: Tracking and Routing
• Real Scenarios and Experimental Results
• Conclusions and Future Work
![Page 3: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/3.jpg)
What is Navigation?
3
1) Determine your location
2) Indicate the route
SOUTH
![Page 4: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/4.jpg)
Global Positioning System (GPS)
4
Provides positioning and navigation services
There are currently 72 satellites (Feb 2016)
We need to receive signals from at least 3-4 satellites
GPS principle
![Page 5: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/5.jpg)
GPS: Weak Points
5
The more sky view, the more accuracy one can get
INDOOR NAVIGATIONGPS does not work in buildings
![Page 6: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/6.jpg)
Indoor Navigation: Interesting Use Cases
6
Museum
Airport
High-level parking
Heat map
![Page 7: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/7.jpg)
Possible sources
LTE Direct
WiFiBLE
BeaconsRFID
Micro accuracy use(less than 10 cm)
Incompatible / unsuitable
Marginally suitable / compartible
Compatible / suitable
Close range accuracy (less than 2m)
Medium range accuracy (less than 20m)
Long range use (greater than 20m)
Requires external power
Requres connectivity
Low cost of materials
Geo-magnetic
7
![Page 8: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/8.jpg)
Examples of Beacons
8
BLE – Bluetooth Low Energy
![Page 9: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/9.jpg)
Beacon Parameters and Package StructureName Data format Range
Set UUID 16 BYTE 16bytes of data
Set Major UINT 16 2bytes of data
Set Minor UINT 16 2bytes of data
Set Name character 20 BYTE 20 BYTE
Set contact BOOL NO -YES
Set Broadcast IntervalUINT 16
0-1600
Set measured power UINT8 0~-129
Connect/disconnect BOOL NO~YES
Password paring UNIT 12 “654321”(ASCII)
Temperature and
humidity ON/OFF button 1 Byte 1 BYTE
Temperature and
humidity read1 Byte 1 BYTE
9
...
Proximity UUID
Major 2Major 1
Minor 1 Minor 2 Minor X Minor 1 Minor 2 Minor X
Sky beacons
![Page 10: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/10.jpg)
Indoor Positioning: Fingerprinting
Principle of fingerprinting
Saving of radio map Measurement and estimation
10
![Page 11: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/11.jpg)
Indoor Positioning : Trilateration
11
Mobile phone
Beacon
),( 111 yxR
),( 222 yxR
),( 333 yxR
1d
2d
3d
![Page 12: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/12.jpg)
User Location in Buildings
![Page 13: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/13.jpg)
High-Level Description of User Location Algorithm
12
RSSI
measurements
Distances
calculation
Location
estimation
Building Map Beacons locations
![Page 14: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/14.jpg)
RSSI from Beacons
13
RSSI – received signal strength indicator
Fluctuations are very high
RSSI is sensitive to environment
Example of RSSI from 3 beacons
Sensoro 4AA
![Page 15: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/15.jpg)
𝑃𝑟 ∝1
𝑑2
𝑃𝑟 =𝑃(𝑑0)
𝑑/𝑑0𝑛
Range vs RSSI: Propagation Models
14
All electromagnetic waves demonstrate inverse-square relationship
between RSSI and distance
Difference between transmitted and received power is described via
“path loss”
Typically d0 is set to 1m
higher RSSI not always better
[1] Indoor Location Tracking Using Received Signal Strength Indicator
By Chuan-Chin Pu, Chuan-Hsian Pu and Hoon-Jae Lee DOI: 10.5772/10518
Parameter n depends on environment
RSSI vs distance
![Page 16: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/16.jpg)
if ( == 3 ) {
doTrilateration();
}
if (
doMultilateration();
}
> 3 ) {
Multilateration
15
![Page 17: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/17.jpg)
Application of Trilateration for 3
beacon signals with the greatest powerApplication of Multilateration
Trilateration vs Multilateration (Simulation)
16
![Page 18: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/18.jpg)
Calibration Procedure
1. tx_power = measurement of rssi
on distance of 1m from a beacon
→ obtaining a statistical sample
→ averaging.
2. Path loss exponent n = rssi
measurement at several fixed
distances from a beacon →
comparison with the theoretical
values → calculation of n to
“adjust” the calibration curve to
experimental data
Example of adjusted calibration curves (rssi is
measured at 1.5, 3 and 5 m from each beacon)
Beacons are different
17
Calibration principle
![Page 19: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/19.jpg)
Calibration Effect𝜎𝑥 𝑐𝑎𝑙𝑏𝑟 = 1.6472 𝜎𝑥 𝑢𝑛𝑐𝑎𝑙𝑏𝑟 = 10.2921
𝜎𝑦 𝑐𝑎𝑙𝑏𝑟 = 0.8888 𝜎𝑦 𝑢𝑛𝑐𝑎𝑙𝑏𝑟 = 6.1412
Estimated user location (static position)18
Map
![Page 20: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/20.jpg)
Effects of errors in beacon
coordinates up to about:
±0,3 m (case a),
±1 m (case b),
±1.5 m (case c)
a) b)
c)
Effect of Beacon Location Error (Simulation)
19
![Page 21: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/21.jpg)
Indoor Navigation: Tracking and Routing
![Page 22: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/22.jpg)
Indoor Navigation:Required Inputs
Building Map
Inputs
Configuration data
Beacons locations
Allowed zones
Building Map
We need to know locations of beacons
User should know only about map!
Black box
20
![Page 23: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/23.jpg)
Building Map and Mask
We can mark the allowed areas with simple mask
Possible user locations are determined
Building map Binary mask
21
![Page 24: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/24.jpg)
+ =
Map and Graph for Routing
Building map Binary maskMap + Mask
Building map
Simple graph example
22
![Page 25: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/25.jpg)
Mask Skeletonization and Graph Extraction
Mask skeleton
23
Mask
Thinning algorithm can be applied
Graph is easily obtained from skeleton
Extracted graph
In the case of large maps we really can
save time
![Page 26: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/26.jpg)
Geodesic Distance
Binary mask
Route example 1
Route example 2
Route example 3 Shortest path can be extracted directly from the mask!
Geodesic distance is determined as a shortest distance
between the given pair of pixels within the image mask
24
Geodesic distance
![Page 27: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/27.jpg)
getRotationMatrix(gData,
mData)
Get device angles with
respect to the ECS
gData – Gravity sensor data
mData – Magnetometer data
return – Rotation Matrix
getOrientation(RMatrix)
return – Roll, Pitch and
Azimuth angles
Mobile Sensors and Orientation
Device orientation Earth coordinate system
25
![Page 28: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/28.jpg)
Magnetometer sensor data Filtering effect
Filtering of Sensor Data Example
Fluctuations in sensors data affect on data augmentation quality
Sensors are crucial for user navigation
26
![Page 29: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/29.jpg)
![Page 30: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/30.jpg)
Real Scenarios and Experimental Results
![Page 31: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/31.jpg)
System Setup
Mobile SDK Indoor map
Mobile phone (client)
Beacons locations
Server
1) User installs SDK
2) Map and configuration data are loaded from server
3) We are ready!
27
![Page 32: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/32.jpg)
• Developed by Apple.
• Has deep integration with the iOS
Works out of the box with CoreLocation
framework.
• Broadcasts 1 advertising packet
(UUID, Major and Minor)
• Developed by Google and open to
extension
• Works on Android and iOS.
• There are a lot of manufactures and
SDKs available for both mobile
platforms.
• Broadcasts 3 advertising packets
(frame types):
• Eddystone-UID (similar to iBeacon)
• Eddystone-URL
• Eddystone-TLM
iBeacon and Eddystone
28
![Page 33: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/33.jpg)
Configuration Tool
29
Building map in tool Control panel
![Page 34: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/34.jpg)
Map and Stream Modes
Map mode Stream mode
30
![Page 35: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/35.jpg)
Conclusions and Future Work
Indoor navigation is very promising direction
Beacons are low cost and easy to use
SLAM technologies
Accuracy improvements
Proper signal processing allows to achieve a competitive
localization accuracy
System performance optimization
Areas of application are extensive and the market is growing
![Page 36: Advanced Indoor Navigation - AI Ukraine · Advanced Indoor Navigation I.M. Gorovyi CEO at It-Jim, PhD, signal and image processing. OUTLINE • Introduction to Navigation • User](https://reader033.vdocuments.net/reader033/viewer/2022042305/5ed0900dbab34719bb2d1738/html5/thumbnails/36.jpg)