Bringing Navigation Indoors

Transcription

1 Bringing Navigation Indoors Fabio Belloni Principal Researcher NRC Radio Systems Laboratory Finland

2 Contents Why going indoors? Use cases, opportunities, and challenges Cognitive Positioning Hybrid positioning systems Indoor Map meets Positioning Technologies the Positioning Content Layers Angular-based Positioning System High Accuracy Indoor Positioning (HAIP) Example of hybrid positioning system Combining IMU, HAIP, and indoor maps BLE proximity Using standard Bluetooth 4.0 for positioning Come to see the demos! 2

3 Why going indoors? 3

4 Motivation for Indoor Positioning Location based services (LBS) are growing enabled by mobile devices with GPS; e.g. Navigation and guidance Sport, training, and health Social networks Security and Emergencies Accessibility People spend most of their time indoors No wide-spread indoor positioning systems and services are available yet GPS operation is very limited indoors Personal communication devices can enable indoor positioning through local wireless networks Big market opportunity in indoor LBS Time spent People spend 80-90% of their time indoors 70% of cellular calls and 80% of data connections originate from indoors. (Source Strategy Analytics) 4

5 Example of Consumer Needs B2C use cases Where to find milk? Who s around? Where did I leave my car? Where is the closest restaurant? How to get there? What can I find here? 5

6 Example of Businesses Needs B2B use cases Is real-time security management possible? Retail chain, Mall manager, Service Providers, System integrators, Marketing analyst, Car manufactures, etc. Is the layout of my store effective? Can I extend the onboard car navigation? How can I attract customers to my business? I want to track my goods? I want to build a dynamic advertisement platform 6

7 ...mentioning some companies looking at indoors 7 SkyHook Wireless (USA) Qubulus (Sweden) Ekahau (Finland) GloPos (Finland) PointInside (USA) WLAN Tracker (Germany) SenseWhere (UK) SEER Technology (USA) FootPath (Germany) Roodin (Italy) Teldio (Canada) Locata (Australia) OmniSense (U.K.) BatPhone (USA) Q-track (USA) Motorola Google Microsoft Samsung Apple RIM Nokia Sony Ericsson...the list is not complete and in no particular order...

8 Indoor environments are very different Open/close spaces High/low ceiling Static/dynamic Metal surfaces Crowded/empty 8

9 Indoor environments are in 3D (.or 2.5D) Multi-floors Floors opening Connection points; elevator, escalators, staircases, etc Usually: 3D: Lat/Lon/Alt 2.5D: Lat/Lon/Floor (or level) Overall Challenge: How to educate the market! For success: having the solid technology solution is necessary.. but it may not be sufficient! 9

10 Cognitive Positioning 10

11 Exploiting different Positioning Technologies GNSS/AGPS WiFi BT/BLE Fusion filter Application IMU Maps 11

12 Mobile Devices using Hybrid Positioning Positioning Technologies need to be transparent to users. The device need to become positioning aware and switch between positioning technologies seamlessly. Proper handling of transition and handover areas is crucial For example: the Application sets the QoS requirements The Filter select the most appropriate positioning technology: meet the requirements by using the least power consumption Data Fusion could be carried out using Hard decisions and switching between technologies Particle Filter Kalman Filter In order to combine independent data streams, it is very important that each estimation technologies provides at least the real-time variance (or any measure of uncertainty) of the generated location data Some form of Time Stamp 12

13 Indoor Maps & Positioning Content Layers 13

14 Positioning Content Layer Maps comprises of several layers of information, having different roles: For visualization and rendering Search Routing Navigation and guidance The Positioning Content Layer (PCL) comprises of information needed for running the positioning algorithm. PCL are invisible to the user. In order to make the whole indoor solution scalable, secure, and sustainable, the PCL need to be referred to one map This should also simply synchronization and data management 14

15 Positioning Content Layer Classes of positioning technologies WLAN IMU RFID BT MAP BLE 15 Finger Printing AP location Radio Maps Info used for data filtering Location of RFID tags Attributes (e.g. ID, type, etc) Location of BT tags Attributes (e.g. ID, type, etc) Calibration data Locations of anchor nodes Attributes (e.g. ID, type, MAC addr, Tx channel, etc) Calibration data Layer for map rendering and visualization Layer with the POI and their attributes Raw map data Layers with information visible to the user Positioning Content Layers Info enabling specific positioning algorithm

16 HAIP High Accuracy Indoor Positioning 16

17 HAIP Location Enhanced Bluetooth Technology Built on top of Bluetooth Low Energy (BLE) technology a new protocol allowing direction finding capability has been added Provides up to 0.3 m position accuracy with <1 sec delays Based on directional positioning beacons installed in covered areas No calibration of the radio environment required HAIP can operate in two configuration modes Assets Tracking (or network centric) Mobile Centric HAIP technologies is based on measuring angular estimation, from or to the locator. The locator is a switched antenna array multiple antenna elements and one BLE transceiver. 17

18 Two Modes of HAIP Technical Principle Network centric mode: A Tag or a Mobile Phone transmits BLE packets at regular intervals Locator(s) receive(s) the packet by using a switched antenna array and provides measurements to a centralized localization server Mobile centric mode: Locator(s) transmit(s) BLE packets at regular intervals by using a switched antenna array A Mobile Phone receives the packet(s) and calculates its own position Locator TAG Phone AoD Locator RF switch BLE TRX AoA BLE TRX BLE TRX RF switch BLE TRX AoA = Angle of Arrival AoD = Angle of Departure 18

19 Localization Principle with a Single Locator z TX / RX Using a single antenna and fixed mobile height, mobile can resolve its 2D location θ z m h m = m h m y m y φ RX / TX x m x TX: transmitter RX: receiver 19

20 Localization Principle with Multiple Locators z z \ TX / RX TX / RX Using multiple positioning beacons, mobile can resolve its 3D location or increasing the position reliability and accuracy z m θ 1 y m θ 2 φ 1 RX / TX φ 2 x m x x \ TX: transmitter RX: receiver 20

21 Server-Centric configuration Positioning Server Locator (in receive mode) 3GPP WLAN Map + PCL BLE Tag BLE phone 21

22 Mobile-Centric configuration One way communication is sufficient between the mobile and the locator; i.e. time synchronization is not needed Locator (in broadcast mode) Map + PCL Bluetooth Low Energy (BLE) 2.4 GHz Broadcast 3GPP WLAN. (network connection) Handset with HAIP onboard 22

23 HAIP Benefits Small power consumption Offers high accuracy; ~1 m in large open indoor areas and 0.3m in office space No calibration required No synchronization required Provides clearly better accuracy and reliability than fingerprinting based WiFi positioning technologies Specific to Server-Centric approach Tags are small, cheap, and have low power consumption Tags can be commanded to become active for real-time tracking Specific to Mobile-Centric approach Beacons may be battery powered and standalone (no network connection required) Allows unlimited number of mobiles to position themselves No network dependence => no cost, no latency, no privacy concern No additional HW required in the phone 23

24 HAIP Experimental results Real path Measured path Positioning Beacon X 24

25 HAIP Demonstrations & Videos 25

26 HAIP Technology Status Current prototypes are built by modifying the Bluetooth Low Energy standard Discussions initiated late 2010 about standardizing the technology as part of future versions of Bluetooth BT SIG as standardization proceeding Stage 1 passed; i.e. MRD & FRD (Market and Feature Requirements Document) Technical standardization work on going in 2012 Estimated availability of chipsets starting in 2013 onwards, presuming successful standardization Industry consortium under planning to support infrastructure ecosystem creation Verification of new business opportunities and creation of joint industry solution Pre-commercial pilots during 2012 being planned by Nokia No information currently available on product schedules 26

27 Example of hybrid positioning system 27

28 Pedestrian Dead Reckoning TARGET: to evaluate the performance of a Nokia made sensorbox for PDR Nokia sensor boxes were calibrated and characterized against commercial MicroStrain IMUs Plain PDR relies on step detection and requires step length calibration for the user But: in hybrid positioning case step length calibration can be handled automatically 28

29 Motivation Solutions for outdoor positioning do not directly apply indoors, and other easily deployable approaches are needed MicroStrain 3DM-GX2 With development of MEMS technology and sensors inertial navigation show some promises No infra required fill the gaps Through inertial navigation the distance and heading of the user is tracked from a known initial position and direction therefore inertial navigation alone will not suffice, but complementary aids are needed 29

30 Approach Complementary Kalman filter *) tracks the error states of 15 parameters: position, velocity, attitude, gyro and accelerometer biases (all in 3 dimensions) Map information can be taken into account through particle filter Pros: No need for separate step calibration, but it s ready to go Cons: Computational costs *) See details in E.Foxlin: Pedestrian tracking with shoe-mounted inertial sensor, Proc. IEEE Comp. Graphics,

31 Results Only inertial navigation, no map info used Initial position and direction of motion needed Begin & End End Begin 31

32 Results Climbing up the staircase Only inertial navigation, no map info used 5th floor 4th floor 3rd floor 2 nd floor 1st floor 32

33 Fusion filter and hybrid positioning system TARGET: improved accuracy, reliability and reduced complexity of calculations of hybrid positioning system 33 PDR + HAIP + Fusion filter running on PC

34 An example of the particle initialization Angle-based location probability matrix covering the room where multi-antenna array is located. The corresponding particle distribution within the room. 34

35 Fusion filter 35

36 Example results Fusion of map and PDR in cafeteria PDR route clearly deviates, but it is corrected by fusion filter very detailed location and route shape information in HAIP enabled area thanks to sensor PDR route Full fusion of map, HAIP and PDR around auditorium 36 In errorless PDR case this point should be (0,0): now there is about 8 m error after walking around auditorium

37 Example Results Using map data in the fusion filter helps improving data quality, at least in a dense indoor environment 37

38 BLE proximity Positioning 38

39 BLE Proximity Positioning Concept description Applicable on BLE devices: with BLE tags: Tags put on known positions and used as anchor nodes Only tag address + RSSI is used, data payload arbitrary Any BT or BLE device usable Tags used as anchor nodes; e.g. Place one in each room The appplication runs on Nokia N9 devices and it uses the BLE chip already on board of the device to detect the HAIP BLE tags Different algoritms for positioning Tag with strongest RSSI (room environment) Interpolation for multiple tags (open space) 39

40 Beneficts The system uses standard Bluetooth 4.0 technology Tags time life 1-2 years, practically no maintenance needed Easy installation They can be hidden without altering the appearance of the indoor environment No cabling needed, not even power cable Effectively, their installation is as simple as install fire alarms NO system maintenance The positioning system does not need to be calibrated NO Finger Printing data and/or radio maps (as for WLAN positioning) needed Nokia Research has develop a simple tool allowing anyone to deploy and set up its own BLE proximity positioning system. Part of this work is done in cooperation with NAVTEQ Research where also integration to indoor maps is taken into account 40

41 Thank You! F. Belloni, V. Ranki, A. Kainulainen, and A. Richter, "Angle-based Indoor Positioning System for Open Indoor Environments", Proceeding of Workshop on Positioning, Navigation and Communication (WPNC), Hannover, Germany, P. Kemppi, J. Pajunen, V. Ranki, F. Belloni, T. Rautiainen, Hybrid positioning system combining angle-based localization, pedestrian dead reckoning, and map-filtering, International Conf. on Indoor Positioning and Indoor Navigation (IPIN), Zurich, Switzerland, P. Kemppi, J. Pajunen, T. Rautiainen, Use of Artificial Magnetic Anomalies in Indoor Pedestrian Navigation, Vehicular Technology Conference Fall (VTC 2010-Fall), Ottawa, Canada, 2010 Link to video demonstration of HAIP: HAIP in 3D configuration: 41