A Distributed Indoor Navigation System Leaving the Trails Thomas Wießflecker Terena Networking Conference 2009

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009  Everyday life in buildings  Complex real estates  Need for effective disaster management Current situation in cities

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Why do we need indoor positioning? Difficult overview and access in case of hazard Difficult to protect against terror Favourite target Complicated navigation in buildings Increase building security Save time and costs Increase consumer satisfaction

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Overview Introduction Existing indoor positioning systems A new approach: Computer Aided Disaster Management System (CADMS) Building Information Model (BIM) Indoor map matching Evaluation Conclusions

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Why do we need to leave the trail? Existing positioning solutions 1.GPS cannot be deployed inside buildings 2.Systems based on RFID, WiFi, UWB depend on stationary infrastructure  hardly adaptive, cost-intensive  not suitable during an incident 3.Current map matching techniques cannot be deployed inside buildings Development of a totally new approach at Graz, University of Technology Main benefit of the CADMS: Autonomous indoor positioning

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Approach to functioning indoor pos. Development of a Computer Aided Disaster Management System (CADMS) –Collecting acceleration data with Inertial Measurement Units –Sophisticated graphical user interface –Transmission of information to remote service customers –Integration of imprecise sensor data and static data describing the real estate Development of a consistent Building Information Model (BIM) Embedding of dynamic sensor data into BIM data  Increase in positioning accuracy

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Deployed hardware Inertial Measurement Unit (IMU) Head mounted display

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 System architecture Update Engine data storing local position data remote position data status information network floor plans dwg, dxf, atl standardised XML format signalling status information corrected position data additional building information CAFM Real- time verification and correction communication graphic display within a lightweight application running in the control room and on-site

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 The BIM’s requirements Represent up-to-date information about the real estate (digital floor plans, room information, room polygons, area information, doors, stairs, reference points, columns, shells, dangerous goods etc.) Allow for presenting this information in a context-adaptive way  include capabilities for indoor positioning Context-adaptive BIM that also integrates sensor data and structural information, adaptable to events occurring during the deployment

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 BIM: Accessibility Model real estate building floor room transition area reference point dangerous goods if located outside the building area transition reference point dangerous goods

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Map matching for road traffic true position corrected position measured position P t Set of acutal streets N Set of estimated arcs Goals: Determine the arc P t should be associated with Determine the street is currently on Determine the exact position on this street

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Again: Leaving the trails Digital representations of buildings Room Door ? Window Vain attempts to model rooms and buildings by using arcs For the CADMS rooms are represented by polygons and doors by transition objects forming a 3D building model.  New indoor map matching strategies are desperately needed! ? ?

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Indoor positioning Improvement of plausibility and accuracy by integration of inertial measurement data and BIM data Automatic repositioning –Innovative map matching algorithms taking the geometry of the building into account Manual repositioning –Correction of the displayed position by user interaction

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Position determination evaluation of physical correctness plausibility check and automatic repositioning manual repositioning area determination height calculation plausibility level calculation adjusted position history sensor measurement data user plausibility level information stored in the BIM display

Decisions and subsystems for automatic repositioning Applied for each position measured by the IMU

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Automatic repositioning in the domain of a door Only physically possible routes through the building are allowed to be covered. Drifting measurement results are adapted and corrected.

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Door correction current Room next Room 1’ 2’ intersection point trajectory - wall door barycentre 2 1.Determine the intersecting wall and the corresponding room polygon 2.Derive the intersection point of the polygon and 3.Move towards the door barycentre 4.Adapt the length of responding to the room’s geometry 5.Pass on point 2 up to the next layer of the CADMS

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Stair correction height [m] time [s] stair correction performed z-coordinate of the raw positions z-coordinate derived by superimposing BIM data covered route includes a stair calculation of the corrected position

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Stair correction – result

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Manual repositioning Adaptation of measurement results by user interaction and short commands Repositioning (with reference points) –Changing the room –Changing the floor –Entering/leaving the building –Entering clearly identifiable areas: stairs, doors etc. –Communicating the room number of the room currently staying in –…

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Manual repositioning (examples) Determine the absolute position with a reference point (door object) “door on the left!”, “door on the right!” Belated correction of entering the wrong room on the digital floor plan on the head mounted display “other room!”

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Qualitative evaluation (1) Uncorrected sensor data Physical layer: IMU mounted on the pedestrian’s torso

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Qualitative evaluation (2) Indoor map matching exploiting the BIM Physical layer: IMU mounted on the pedestrian’s torso first floor

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Qualitative evaluation (3) Indoor map matching exploiting the BIM Physical layer: IMU mounted on the pedestrian’s torso second floor : manual correction

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Qualitative evaluation (4) Physical layer: IMU mounted on the pedestrian’s foot Uncorrected sensor data Indoor map matching

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Qualitative evaluation (5) Physical layer: IMU mounted on the pedestrian’s foot Uncorrected sensor dataIndoor map matching

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Outcomes and conclusions (1) The measurement results reveal that the developed approach to providing seamless autonomous indoor positioning can function in practice. This is shown by using two different sensor subsystems and working with different building geometries.

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Outcomes and conclusions (2) The combination of static BIM data and dynamically generated sensor data improves the accuracy of indoor positioning significantly. Providing more intelligent algorithms for the automatic repositioning helps avoiding time-consuming user interaction. A trade-off between intensive building modelling and greater positioning accuracy has to be made.

Indoor Navigation Leaving the Trails Thomas WießfleckerJune 11 th, 2009 Future work Refinement of algorithms for the automatic repositioning techniques Standardisation of the BIM’s structure Optimisation of the user interface Tests during regular fire practices Et cetera

Thank you for your attention!