First, sorry for the slow past 2 months, I’ve been heads-down working on some new mobile enterprise stuff as an EIR at SRI. If you want to trade some notes on enterprise, feel free to give me a ping.

Second, I had recently moderated a WCA panel on in-door mapping. Fascinating discussion from all different angles and I thought to best explain what I learned was in the table below. Note, I’m definitely not an expert in this area and so feel free to comment if I made any errors and/or I am missing any important data.

ServicesOut-door has created a few killer services. First, routing, driving directions and real-time navigation has an entire ecosystem (eg Networks in Motion which powers VZ Navigator or Telenav's in-car navigation solution). Second, local search, which continues to be a larger and larger part of all mobile search. And finally, the mashup ecosystem: Google's making of the Google Maps API available as a free API (where previously you had to pay, for such access) enabled an entire ecosystem of interesting data visualizations. Unlike the out-door, in-door has not yet found it's killer service (or as it seemed in the panel). Some use cases include hospitals, museums, corporate campuses, malls, airports, events and more. Companies such as Presdo, Spotlight Mobile, Fastmall, Geodelic, Ekahau and others are all focused on different verticals / segments. What's most interesting is that the primary use case in each of these verticals is different, yielding almost a vertical-specific in-door mapping experience (and associated business model). For example, museums use in-door maps as a virtual guide but malls use in-door maps to push advertising, promotions and coupons etc.
POI (Points of InterestOut-door POI has mostly been aggregated from directories (eg Whitepages) and crowd-sourced by end-users. Whereas POI was once considered the differentiator, it's now mostly publicly available (eg see Cloudmade)In-door POI is a different situation - like in-door map data, there are POI data privacy issues and the types of POI can be vary significantly depending on the use case. For example, hospitals may want to track assets (such as wheelchairs) but companies may want to track the location of colleagues? Some POI like conference rooms may be fixed but other POI like packaged goods in a grocery store are constantly being moved to different shelves.
PositioningGPS, Cell-ID, network lookup and WiFi are all different ways to determine location. With the smartphone explosion, GPS is becoming the standard and thus why all map data in the out-doors must be geo-coded (so you can be located by satellite etc). Numerous traditional infrastructure companies like Telogis and Polaris Wireless (E911 for network lookup) and many others have enabled a robust out-door positioning capability.Unlike the out-door, GPS is often ineffective in the in-door (and thus why in-door map data doesn't necessarily need to be geo-coded). Walls create barriers resulting in inaccurate GPS positioning and/or an absurdly long time-to-first-fix. As a result, most in-door positioning has been focused around WiFi (eg Skyhook, Cisco, Qualcomm etc). Cisco and Qualcomm in particular, have been investing significant money in improving their WiFi chipsets to enable better positioning (aiming for 3 meters, instead of 6 meters) but you still have the challenges of multi-story buildings etc. Additionally, there has been some clever exploration in the use of the camera (eg augmented reality) and/or the use of the pedometer to determine distance from a start point and then location. Similar to the out-door, in-door location will be a blend and heuristic of multiple technologies.

Knowing that, the bigger question is whether the in-door requires positioning? Certainly, there are some applications where positioning could be useful (eg navigating a warehouse), there are many applications where positioning is a nice to have (eg navigating an airport where you can really only go left or right). Since the in-door has never had positioning, the venues tend to have numerous signs/arrows/painted lines etc (eg hospitals draw colored lines on the floor leading to different rooms) assisting the user.
Map RenderingRendering (with great UX) is what allowed Google Maps to surpass Mapquest (and all of them have since improved). With out-door, you have numerous open-source packages (eg Mapnik, OpenLayers etc) and commercial vendors like Decarta (apologies, if I'm not including all of them).In-door maps can obviously be rendered with the same open-source packages available for the out-door but the we have a couple other things to think about. Out-door traditionally doesn't have to deal with the Z-axis (eg for multi-story buildings) and more importantly, it's not clear if the in-door should be rendered in the same user interface as the out-door? Whereas the out-door has arguably fewer POI (points of interest) per square meter, if you are rendering an in-door map for a grocery store, the volume of POI is significant and thus the way you may need to think about rendering the map may be quite different. Additionally, unlike the out-door which has been fairly uniform (eg roads, buildings etc), in-door may require vertical UX for different applications (eg navigating a museum versus a park versus a mall versus an office building will be different).
Data CollectionNavteq and Teleatlas have their roots here. Google, Bing followed for strategic reasons. And numerous crowd-sourced solutions followed like Waze and Open Street Maps. Most of these folks collect the data (using vehicles) and then make the data available in a standard map data format such as a GeoJSON or KML. In most instances, this data is geo-coded (mapped to a real lat/long) for positioning purposes.The In-door data collection process isn't too different from the out-door and there are some new startups collecting this data. Micello, Point Inside seem to market themselves in data collection (eg touting their volume of maps). The big boys are also playing here - Navteq has a destination mapping group (interesting name) and Google recently announced trikes (3-wheeled bikes) to pedal through parks, outdoor malls etc (not really in-door) but definitely getting there. The big difference though, is that in-door data is not readily available and cannot easily be crowd-sourced. Most in-door startups today are partnering with venues to acquire this map data and/or literally photographing fire-escape maps or printed wall maps etc. In-door data unlike the out-door may have specific privacy issues as well (eg can my in-door maps be visible to everyone or just my team?)

Not talked about in the table (but only alluded to) is business models. Out-door is clearly monetizing local search but in-door is still unclear? On the panel, this was the one question along with clever tricks for in-door positioning that nobody opted to answer :)

Update: Just saw this partial video of the WCA event posted on YouTube:

Tagged with:

7 Responses to In-Door Mapping

  1. Anil Agarwal says:

    Great summary, although we at Micello can summarize in fewer words – Outdoor is all about maps for car, indoor is about the maps for people!
    To answer your question on tricks for indoor positioning – from the companies we have tracked –
    1. WiFi – Fingerprinting
    2. WiFi – Signal strengths measurement and triangulation
    3. Long wave TV signal
    4. Cell tower signal chatter
    5. Audio signal
    6. Bluetooth & Zigbee
    7. MEMs

    I am sure there are still more tricks under the cover.

  2. Matt Miesnieks says:

    we did a *lot* of research on various positioning options at Layar. There are a few that weren’t mentioned above that are probably going to end up most-promising IMO:
    – point cloud based determination of position (see earthmine. google is also collecting LIDAR pointclouds now too)
    – Ultra Wideband solutions (similar to wifi based positioning, but millimeter accurate. all solutions are proprietary today)

    The industry is in general agreement that a combination of solutions will be adopted, as each as pros/cons in different environments. Getting the perfect cocktail will take a while.

  3. Raj says:

    Thanks Anil, I much prefer your summary :)

    Yes, there is a laundry list of positioning techniques – fascinating to see those Trimble backpacks using TV RF to determine location, impressive.

  4. Verizon Wireless provides free WiFi geolocation service as part of the free NavBuilder LBS SDK. which is cross device for Android , RIM, Brew , Windows ..

  5. Bryan Wargo says:

    I believe retail is going to be the market with the greatest need for indoor positioning, especially the big-box stores. Everyone is looking for a solution that provides 3 feet of accuracy without requiring a whole new set of infrastructure. The ROI comes from manufacturers who are willing to pay to get access to those shoppers when they are actually next to the products they make. The hard part is actually making the technology and ecosystem work.

  6. Frank Schuil says:

    Hi Anil,

    Good summary and sharp performance in the panel. Nice to see fingerprinting on top og your list ;)

    Cheers, Frank

  7. Carl Reed says:

    Nice summary! I did not attend the meeting but very much appreciated the concise but informative summary. The Indoor location and navigation domain is receiving considerable attention in multiple fora these days. The following information adds to the mix of activity and information.

    I work for the Open Geospatial Consortium (OGC), a voluntary consensus standards organization focused on interface, encoding, and API standards work that enables the integration and use of geospatial (and spatial) content and services into any application or workflow.

    Since the middle of 2009, the OGC Members have driven an increasing amount of standards activity related to the requirements for Indoor applications. Our work assumes that there is exists a mechanism for provision of location data and that communication of that location to some device or application is possible. The OGC focus is more on providing standard interfaces and encodings to communicate the location data as well as interfaces and encodings to support applications such as tracking, navigation, sensor fusion, indoor mapping, floor plans and so on. From this perspective, we are coordinating with various other standards organizations that are also working the indoor space. This collaboration activity includes work with the IETF, W3C, OASIS, and ISO. There are numerous requirements that have been documented, such as how to seamlessly move from an outdoor reference system into an indoor reference system. One cannot assume that an indoor application will be using a WGS 84 geoid in 2d expressed as latitude/longitude coordinates! This requirement is critical to many emergency response applications. In terms of a “killer app”, well, if one listens to the emergency response community the indoor focus is on first responder safety, citizen evacuation, and other applications that save lives and reduce impacts. Obviously, content collected for consumer applications would be critical for use in EM situations and hence the requirement to be able to easy share content when and where needed is critical.

    In terms of the work of the OGC and our partners, below are some references that might be of interest:

    OGC summit on Indoor Location, Navigation, and Floor plans

    OGC Engineering Report on Indoor Navigation

    OGC overview of Indoor Location focus

    Requirements and Space-Event Modeling for Indoor Navigation
    There is considerable work on “InDoorML” in the OGC. This work is currently being driven by the German, Korean, and Taiwan communities and is related to something known as u-City.

    Related to the work of the OGC, the IETF GeoPRIV Working Group has had numerous discussions and standards development work focused on indoor location and navigation. An interesting set of requirements is related to relative navigation as opposed to absolute navigation. An example of relative navigation is “Take elevator 1 to the 4th floor, turn right and go three offices on the left to room 406”. There is an internet draft describing this work.

    Any questions, please let me know!

    Carl Reed

Leave a Reply

Your email address will not be published. Required fields are marked *