Location, location, location! The proliferation of indoor positioning and what it means and doesn’t mean for museums

Matthew Tarr, American Museum of Natural History, USA

Abstract

Keywords: location, context, wayfinding, personalization, user experience

1. Introduction

Many venues are in the process of adopting indoor positioning systems and wayfinding solutions, incorporating them into mobile apps, and hoping to improve visitor experience while gathering meaningful analytics. Industry giants like Google, Microsoft, and Apple are moving into the space (on the heels of a deluge of startups), mapping civic centers, hospitals, airports, stadiums, amusement parks, and even museums. With a clear view of the sky and a smartphone, there is no longer any reason to ask directions. Why shouldn’t this same device provide us with necessary location-based information during the “90 percent of our lives that are spent indoors”? (Costa, 2013) “Indoor location” is on track to become commonplace, but that doesn’t make it easy.

In this paper, I will discuss several key aspects of “indoor location” as it relates to museums:

• What is location?
• What concepts, technologies, and methodologies inform and support indoor location?
• Locations as one part of context: the future of content delivery.
• Aside from wayfinding and content delivery, what does location mean to museums?

2. Why is location so important?

Like many cultural institutions, the American Museum of Natural History has grown over the years. When Ulysses S. Grant laid the cornerstone for the first Museum building at the 77th Street site, he might not have guessed that nearly 150 years later the Museum would be made up of more than two dozen interconnected buildings. This “interconnectedness” allows the uninterrupted meandering from the Big Bang to the current efforts to eradicate diseases, and from New York’s forests to the edges of the observable universe. However, the Museum isn’t an easy space to navigate.

This presents a very real, and hopefully solvable, problem for the visitor. In addition to wayfinding signs, Museums have often provided printed floor plans that, depending on the visitor’s familiarity with floor plans, can be quite helpful. Additionally, self-guided tours have been published with varying production values, from leaflets to bound books to audio tours, all in an attempt to provide additional context-specific interpretation. What these static maps cannot provide is any sort of real-time information about where the visitor currently is. I am often approached by lost tourists who look at me pleadingly over the unfolded floor-plan brochure. My first step is always to orient the map (even if that means making it “upside down”) and point out where they would be standing on the map, the equivalent of a “you are here” from fixed maps. I quickly give a general sense of the route and an encouraging (and sometimes euphemistic) description of the distance and send them on their way. A system that could accurately identify a visitor’s location and knew the layout of the Museum could provide such simple directions.

Figure 1: the “blue dot” representing my office at the Museum

3. What is location?

The familiar “blue dot” (often used as a synonym for “location awareness”) isn’t one singular technology. Digital “wayfinding” is made up of at least the following three interrelated components: locations, maps, and routes.

Figure 2: locations	Where’s the subject? The primary factor in a location system is the namesake attribute location. It seems simple but in fact is quite complex. Describing the location of something often means agreeing to a coordinate system, a spatial resolution (distance between points), and a labeling scheme.
Figure 3: maps	What else in in the location? The map will often include one or more visual layers. It serves as the display and often one of the primary input devices, in that a user might tap, drag, or zoom the map to modify the area of focus.
Figure 4: routes	What are the possible pathways? Routes are all possible paths that one might travel, as well as a process to determine the path between two points. A path is made up of passable segments. Individual segments may have any number of attributes including length, direction, and quality. Common concerns with indoor routing include indicating segments that adhere to ADA (Americans with Disabilities Act) requirements.

Not all features on a map are visible. In fact, the true map is often a data set of features and shapes. Maps, especially the underlying data, can become out of date quickly and require maintenance.

Of course, these must all share a common coordinate system (lat/long, x/y/z, etc). Many of the offerings coming to market today combine minimal versions of these three layers into a convenient package, often at the expense of advanced features and customization. Historically, users of mapping/location software (geographical information systems) would utilize separate systems or packages for these three distinct functions. Location would be provided by a hardware device using GPS and other detection methods like LiDAR (“laser radar”); maps would be either created or licensed (often at great cost) from private companies or government organizations; and routing would be handled by a software package like ESRI’s RouteSmart for ArcGIS. This complexity has only recently been abstracted by Garmin, Nuvi, and more recently (and effectively), Google, Apple, and Microsoft, to become a “one-click” install and thus an everyday activity.

How do small and mid-size cultural organizations avail themselves of this technological capability? The good news is that there is a lot of attention being paid to this problem, and thus no shortage of options. The bad news is that there isn’t a standout winner.

4. The state of indoor positioning technology

In the 1970s, the U.S. Department of Defense began a two-decade-long, $10 billion effort to provide accurate location to various Earth-bound military systems using twenty-four satellites with highly accurate atomic clocks. These satellites are in geosynchronous orbit, meaning that they orbit at the same speed that Earth rotates. This orbit is also referred to as “geostationary,” as from Earth’s surface they appear to be “stationary.” Each of these twenty-four satellites is a radio beacon. As long as a device can detect two of these signals, it can triangulate a position with useful accuracy.

This elaborate and costly system is provided free for use and is the engine behind the success of in-car navigation systems and Google Maps driving directions. Sadly, the signals from GPS satellites aren’t powerful enough to penetrate buildings, nor is the accuracy afforded by the system fine-grained enough to serve the needs of museum visitors (or mall shoppers, airport travelers, etc). New technologies and methodologies are being created all the time, but how can they determine the signal in all this noise?

Figure 5: indoor tracking – enabling techniques

The value of indoor RTLS (real-time location systems) has prompted some industries to invest the effort (and money) to solve the problem. Hospitals, manufacturers, supply-chain specialists, and perhaps most visibly shipping companies are using indoor positioning to great effect (Hung, 2014). These costly, complicated, and proprietary systems don’t lend themselves to refitting into consumer-facing, cultural spaces.

There is a growing field of candidate IPS (indoor positioning systems) methodologies, including various local, low-power RF (radio frequency) technologies, specifically Wi-Fi, NFC (near-field communication), and most recently BLE (Bluetooth low energy or Bluetooth Smart). Other technologies that are being used, although none with overwhelming success, include audio outside the range of human hearing and light bulbs that “flicker” at a defined rate, much faster than humans can detect, to signal to the device which area it is in. The basic structure is always the same: a signal is generated by a stationary device and sensors on a roving device (mobile phone) senses the signal. These signals are processed with specialized software that attempts to algorithmically turn the signal (and sometimes the relative strength of the signal) into “location.”

Triangulation

While the hardware portion of the indoor positioning battle is still being waged, there is at last some convergence in the software methods being used to determine location. Several types of IPS mimic the GPS system by using stationary RF beacons (Wi-Fi, NFC, BLE) and triangulating the position of a mobile device. If provided enough signal, in both quantity and quality, this method can provide a useful X, Y, and Z (floor) location on a map.

Proximity

The other dominant method for leveraging in-venue signals for location is “proximity.” Popularized by Apple’s iBeacon, this method is able to determine with a high degree of accuracy the distance between a mobile device and a single beacon transmitter. While accurate, the downside of the proximity method is that the “distance” measure could be in any direction. By way of analogy, it is as if a boat is lost at sea and can see only a single lighthouse. It can determine its distance based on the size and brightness of the light, but it cannot tell which direction it approaches from. There are significant dangers when making assumptions based on this incomplete knowledge.

Figure 6: proximity illustrated

Sensor fusion

Regardless of which methodology (or combination of methodologies) is used by a system, the future of indoor positioning will also heavily leverage MEMS (micro-electromechanical systems), also referred to as “sensor fusion.” Sensor fusion is the practice of utilizing all the sensors on mobile phones to augment the location provided by the positioning systems. Currently available sensors include compass, accelerometer, gyroscope, barometer, and an ever-increasing number of newer, smaller, more efficient sensors. In addition, new devices are coming into the market with additional sensors that may be of use to the problem of indoor locations. One excellent example of this might be the constant refining of a phone’s location based on the “number of steps taken” (accelerometer) in a given direction (compass) since the last good location (iBeacon, cell tower, or GPS). These aren’t the only sensors of note; Google’s Project Tango is an experimental product that has multiple cameras, including infrared and stereoscopic lenses, to capture in real time the three-dimensional shape of a room and thus is able to determine the orientation and location of the mobile phone to a very high degree of accuracy.

Mobile-centric versus network-centric

There is another distinction in the software space that is important to understand, although it doesn’t directly impact the ability to determine position. Mobile-centric and network-centric technologies are different only in where the data and/or logic resides for calculating the location of a mobile device. For example, the American Museum of Natural History’s original Explorer mobile wayfinding app used Cisco’s Mobility Services Engine (MSE), which was a device that sat on the network and processed the raw signal-strength data provided by the mobile device to triangulate the location based on a map that was managed and stored in the MSE. The new Explorer still uses triangulation to determine the phone’s location. However, the map and beacon locations are downloaded to the phone, and processing is done locally in the app. The ramifications of this aren’t to be ignored and are most visible in conversations around user privacy and anonymous tracking and reporting.

Why does it matter: Mobile- or network-centric?

This is one of the questions that the public will eventually chime in on. Who should own/control/have access to the secondary and tertiary metadata that can be so meaningful (and valuable and dangerous) in interpreting behavior and actions?

Native location

There is one tiny, bright spot in all this complexity: some big players are trying to solve these very same issues in ways that we will all benefit from. Apple, at the 2014 WWDC (World Wide Developer Conference), announced the extension of Core Location to include the ability to provide indoor maps and location. Almost two years ago, Google began including floor plans on Google Maps for a select set of venues and institutions.

Native location has many benefits, beginning with the phone maker having unfettered access to the phone hardware allowing them to read the radios, and other sensors, to a degree denied to third-party (app store) apps. The other major benefit is that any progress they make in refining the phone’s ability to provide consistent and accurate location will benefit every application developer on the platform (and thus every user of the apps they create).

The very real downside to native location is that there isn’t anything like GPS indoors; instead, they rely primarily on an RF (radio frequency) survey of the venue. Using a smartphone running a specialized app, they actually walk around the venue “listening” with the various antennae for radio signals: Wi-Fi, bluetooth, cellular, etc. This information is then used to create a “map” (or database) of the RF environment at all the various locations. When a visitor uses a smartphone to get a location, the signals detected by the phone are matched against the RF map (and combined with other sensor data) to provide (under optimal conditions) very useful location data. Native location, perhaps unique among all options, will surely continue to improve as smartphone hardware and software advances. Additionally, as native location isn’t dependent on any specific, physical infrastructure (except the RF mapping application), it might see the sort of widespread adoption that can only come from being preinstalled on every device sold.

Figure 7: available indoor position technologies and some qualitative attributes (Otis-Smith, 2014)

5. How and why to choose one technology over others

While Gartner expects to see BLE continue to rise in popularity, the fact that all these technologies are so immature isn’t necessarily a reason to postpone creating a location-aware, contextually appropriate content delivery system. Quite the contrary: their suggestion is to begin looking into these systems “not only for purpose of targeted marketing, but also for customer experience management” (Zimmerman, 2014). Indeed! I would amend that statement to: the user experience (UX) must drive the technology implementation. If the goal is to create sustainable and meaningful experiences, that must come first. The specific technological feature set is in service to the contextual whole.

Late-breaking news and government regulations

Two days before I submitted the final draft of this paper, the FCC announced an update to the E911 (Enhanced 911) rules which require that VOIP (voice over internet protocol) “systems automatically provide a call-back number and in most cases a location.” The updated rules state:

The new rules establish clear and measurable timelines for wireless providers to meet indoor location accuracy benchmarks, both for horizontal and vertical location information. The Commission noted that no single technological approach will solve the challenge of indoor location, and no solution can be implemented overnight. The new requirements therefore enable wireless providers to choose the most effective solutions and allow sufficient time for development of applicable standards, establishment of testing mechanisms, and deployment of new location technology. (FCC, 2015)

The effect this may have on the field of indoor position systems shouldn’t be underestimated. The required accuracy in the new ruling is too low to replace these other options, but the fact that it will (ostensibly) be a standard capability available on all mobile phones is a potent reason to build on whatever the large carriers end up using to get indoor location.

6. Location as UX

It is useful to think about location at three “levels” of increasing refinement:

• Presence: are you at or near the Museum?
• Place: what gallery are you in? Perhaps what section?
• Position: where are you standing, which direction are you facing, what are you looking at?

When expressed in these terms, any content producer can see the value in being able to define/refine the content based on knowing the recipient’s “location.” It is simpler as well to begin to identify what technology is appropriate based on the intended use case. For example, if wayfinding is the primary goal, the “place” is plenty of location specificity for giving directions. At the extreme ends, “presence” is enough to power basic targeted advertisements, and “position” is necessary to provide hyper-specific interpretive media (such as augmented reality).

Location isn’t only wayfinding

Getting from point A to point B isn’t the only value proposition that a smartphone provides from the data its sensors gather. There are endless questions that we are now counting on location-based systems to help answer:

• Is there a four-star burrito south of 18th Street?
• How far is the drive to pick up this free baby seat I found on Craigslist?
• What new movies are playing nearby?
• Remind me when I get home to put the recycling on the curb.
• Are there any cool museum events this weekend within walking distance that are kid friendly?

The idea that saving a step because your phone already knows your location is “cool,” but is it really that hard to type in the location portion of the above queries? While sometimes the answer is yes, the changes that mobile computing have brought into our lives aren’t as simple as having a powerful, location-aware computer with us at all times. If that were true, we’d all be carrying Garmin GPS units instead of smartphones.

Figure 8: tweet

Not a mobile revolution, but an age of context

Location is simply one of the critical aspects of the “context awareness” that visitors are coming to expect. These mobile computers in our pockets, purses, or even on our wrists are able to determine our location, the time of day, our preferences, our history, who and where our friends are, even who are new friends ought to be. All of this could certainly translate to value for the visitor in determining WHAT they are actually trying to find, not just how to get there.

The democratization of access to context-specific content

Last year while researching location technology in San Francisco I stopped by the California Academy of Sciences. Apple had used CalAcademy as a sample location in some of the WWDC presentations regarding indoor location. As a byproduct, it was one of the only places in the world that had been surveyed by Apple. I wanted to experience, firsthand, the updated Core Location indoors! I met up with some colleagues and walked through the building. I love museums. We walked and talked about technology and also about the exhibits. I was given privileged access above and beyond the typical museum visitor. As the wandering progressed, I was struck by how amazing it was to have such access: knowledgeable and responsive voices, behind-the-scenes perspectives, and freedom to follow my curiosity. Why can’t everyone enjoy this level of “access”? Can’t technology be harnessed to increase access to this sort of experience beyond the museum insider and funders to the average museum-goer?

My favorite anecdotal scenario for explaining AMNH’s efforts with the Explorer app and other digitally mediated, interpretive layers speaks to this privileged access. The evening before a new exhibition opens, the Museum president will do a walkthrough. On one side is the scientist/curator who is responsible for the content and overall message of the exhibition, and on her other side is the head of the Exhibition department, the team that has “translated” that curatorial message into exhibits, labels, interactives, and signage for the visitors. She can ask ANYTHING and get an immediate answer. She can be led by the Exhibition’s +You, her own personal interests, or ask a completely tangential question about the technique of mounting a particularly heavy specimen.

Why can’t everyone have this experience?

They can—at least the decidedly down-market version afforded to my personal friends. Having worked at the Museum for many years, if I have the time and am willing to walk them around, I am capable of creating a reasonable facsimile of that experience. #IMHO

Why can’t everyone have this experience (without knowing a museum employee)?

They can, but first we need to encode a LOT of assumptions into the content-delivery processes. Between the curator and the head of the Exhibition department (or just a friendly Museum employee), the following data are available (if not always obvious):

• The spot where she is standing when she asks a certain question
• The object or case she is pointing toward
• Where she has just come from and what she has seen of the exhibition up to now
• The accumulation of previous questions/curiosities
• Her historical understating of the topic
• How late in the day it is
• How late in the tour it is
• How much more time she has to spend in the exhibit
• Some personal preferences (e.g., dislikes penguins, likes snakes)
• The size of her group
• The approximate age of the group members

In some cases, these may not be obvious, and the human “tour guide” can simply ask.

These example factors (contextual indicators) can be logically grouped into the following: location, visitor history, preferences/interests, demographics/group dynamics, and date/time. While there are an infinite number of possible indicators, looking at these will hopefully start to describe a problem-set that can be solved by computers (and sensors).

7. IoT (Internet of Things)

Speaking of sensors, the Internet of Things is a rapidly approaching black hole that I will attempt to orbit, without falling through its event horizon, from whence no rational thought escapes.

Of course by the time this paper is presented, the Apple Watch might be an actual product, and we’ve already had more than a year with Android Wear, NEST learning thermostats, DropCam Wi-Fi-connected baby-monitors, Roomba, and fitness bands that remind you two hours before bed that you still have another 1,100 steps to go. CES 2015, with its significant focus on tiny, smart, sensor-laden gadgets, has faded into memory; some of those products have been released while others never will. Regardless, they all contribute to the mythos around the future of context-appropriate media delivery.

8. Downsides?

Is there a downside to all this contextual customization? How do we embrace context awareness, including predicting visitor needs based on both individual behavior and behavior in aggregate, while balancing those with concerns over privacy?

Let us not ignore the perhaps lesser concerns over the loss of serendipity. As recommendations become better and better, is there a risk that our exposure to novel or unexpected (even unwelcome) ideas will dwindle? #selective_exposure

9. An incomplete picture

While it is critical that we keep up with visitor expectations and begin to provide context-aware services (including location), it is equally important that we keep our eyes on the prize and don’t give ourselves and our visitors a partial picture. In most cases, IPS (indoor positioning systems) will give information and analytics about what the user is doing and where they are doing it at a given time (the proverbial what, where, and when). However, we are all well aware that even the most elegant, thoughtful, and exciting digital experience will only capture a fraction of the total visitorship, and of that fraction only a subset of their actions will be handled via the mobile device. What about the other 75 percent of interactions they have with a museum?

Location as native data point for all interactions

If we include “location” data with every recorded transaction, not only those via the mobile app, we can start to make a complete picture of a visitor’s behavior and preferences.

• Do they always enter through the same lobby?
• Is there a measurable increase in retail activity by visitors exiting via a certain door (between certain hours)?
• Did they buy their tickets within two blocks of the museum?
• Is there a correlation between concentration of visitors on a certain floor and coffee sales on a less busy floor?
• Do “planners” abandon their pre-made tours when they encounter seventy-five screaming elementary school kids in the first hall they visit?

In order to make these determinations, we need to gather “location” information from as many points of interaction as possible, including transactions that happen at a fixed point of sale. Historically, these haven’t been tagged with location. For example, a member buys a hat using her discount (by scanning a physical membership card). We could know on which floor and in which shop she made the purchase. More interestingly, if we track these in the same way we do mobile app interactions (logging them the same way we would a more typical wayfinding event), we can paint a more complete picture of a visitor’s actual day.

The proliferation of location as a source of data to improve both visitor experience and institutional ability to manage its business is only scratching the surface. We need to take a long view on these new data and implement systems that can truly interpret the signal from the noise. Focusing too much on the “easily measured” can lead to new confusions and misplaced efforts. By integrating location (and other context) data gathering as a native attribute in all our analytics, we gain access to a new level of intelligence about our visitors to both meet their expectations and better educate and inspire them.

References

Costa, Tony. (2013). “Next In Tech: Indoor Positioning.” Forrester Research. Available https://www.forrester.com/Next+In+Tech+Indoor+Positioning/fulltext/-/E-RES82781

Federal Communications Commission (FCC). (2015). “FCC Adopts Rules to Help Emergency Responders Better Locate Wireless 911 Callers.” January 29. Available http://www.fcc.gov/document/fcc-adopts-rules-help-responders-better-locate-wireless-911-callers

Hung, Mark. (2014). “Innovation Insight: Indoor Location Technologies – The Looming Battle Between Bluetooth, Wi-Fi and other Wireless Technologies.” Gartner. Available https://www.gartner.com/doc/2753917/innovation-insight-indoor-location-technologies

Otis-Smith, Alexi. (2014). “Wayfinding Recommendations and Analysis.” Door3.

Zimmermann, Annette. (2014). “Indoor Positioning.” In Hype Cycle for Mobile Device Technologies. Gartner. Available https://www.gartner.com/doc/2801235/hype-cycle-mobile-device-technologies