Mappedin started as a school project and evolved into a leading indoor mapping company, working with malls, airports, hospitals, and Fortune 500 companies.

You guessed, today’s podcast is all about indoor mapping, why it’s hard, what are the use cases driving it, what the state of the art looks like today and what we can expect in the future.

Key points discussed include:

1. The Challenge of Indoor Mapping: Unlike outdoor mapping, indoor environments are complex due to the density of objects and frequent changes. Moreover, indoor spaces are mostly private property, making it difficult to collect data comprehensively.
2. Importance of Indoor Maps: Despite the challenges, the need for indoor maps is growing. Applications range from wayfinding in malls and airports to optimizing logistics in warehouses and creating better guest experiences in various venues.
3. Mappedin’s Approach: Enabling non-experts to create and maintain indoor maps. Their tools are designed for everyday users, allowing them to update maps as easily as they would a document in Google Docs.
4. Technological Advances: While technologies like LiDAR and digital twins offer detailed 3D models, are they really practical?
5. Indoor Positioning: Accurate indoor positioning is crucial for the success of indoor maps, similar to how GPS revolutionized outdoor mapping. However, this remains a challenging area due to signal interference and the complexity of indoor spaces.
6. Future Outlook: Digital indoor maps become as ubiquitous as Wi-Fi, providing essential data for various applications and improving overall user experience … but we are not there yet!

Try Mappedin for yourself https://www.mappedin.com/

or connect with Hongwei here https://www.linkedin.com/in/hongweil/

Recommended Listening

• Where does the blue dot come from (how Google knows your location)
• Hyper Accurate Indoor Location
• Using the Geomagnetic fields of buildings to navigate indoors

In Conversation

From School Project to Mapping a Third of the World’s Malls

Daniel: You’re the founder and CEO of Mappedin. Could you give us a brief introduction?

Hongwei: I’m one of the founders here at Mappedin. More than 10 years ago we started this as a school project to help people find stuff indoors — for the local university, the local mall, the local hospital, random buildings who would take our call as a bunch of students. Fast-forward to today: about a third of the malls in the world are Mappedin customers — the nicest mall you’ve been to recently is probably a Mappedin user; you’ve touched the touchscreen, used our web app. The Super Bowl has used us for the last two years, and close to 10% of the Fortune 500’s corporate offices are powered by us behind the scenes for people to find desks, find meeting rooms, view occupancy. We’re one of the largest indoor mapping companies in an emerging space.

How Little of the Indoors Is Mapped

Daniel: Compared to the outdoor world, where a lot has been mapped — where are we with the percentage of indoor environments mapped?

Hongwei: Close to 0% — certainly less than 1%. We’re just scratching the surface. The outdoor world is, for all intents and purposes, mapped — there’s a near-real-time digital map of any place on Earth from 10 different vendors. Indoors, maybe 10% of these premium buildings that have invested in wayfinding and guest experiences have been mapped, but as a whole it’s less than 1%. Most buildings you go into, you’re looking for the PDF — the fire-escape map posted by the entrance by law, the black-and-white thing someone has scribbled on, or a brochure for Disneyland literally drawn by an artist. I wouldn’t call those maps in the way we’ve come to rely on outdoors, and that’s one of the problems we’re trying to solve.

Why Indoor Mapping Is Hard

Daniel: Is the lack of coverage because the problem is too hard, or because it’s not important?

Hongwei: A bit of both, and both have changed in the past couple of years. Back in the early 2000s, DARPA did a study with Esri on what it would take to map the indoors once — they found it would take 10% of the US workforce working 40 hours a week for 10 years. Basically impossible. The indoors is actually denser than the outdoors — once you get into desks and four-foot sectionals in warehouses, there’s so much more to map. And two things make it hard to digitize: first, it’s 99% private property — you can’t drive cars down the road or fly satellites over it; you have to work with hundreds of millions of private landlords. Second, the indoors changes really frequently — airports have swing gates that are international territory in the morning and domestic in the afternoon, Santa moves at the mall every year, desks move, doctors change locations. The indoors is dynamic in a way the outdoors isn’t. And honestly, until recently it hasn’t been that important — outdoor digital maps weren’t that useful until everyone had a smartphone with GPS and a mapping app. Now everyone is talking about digital applications inside buildings, and that’s calling for this dataset to exist.

Enabling the “Paper Scribblers”

Daniel: To do any analysis you need a map first. How are you creating maps today?

Hongwei: We’ve built tools for the average computer-savvy person — not a GIS-savvy or architecturally savvy person. Our goal is to replace the PDF map that is still ubiquitous, and our realization is that we don’t have to do anything miraculous — we just have to enable the paper scribblers. When we started, customers would send us a PDF with scribbles every three months — “here’s what’s wrong, here’s what changed” — and ask us to Photoshop it and update their website. We hated that repetitive work, so we built tools for ourselves: collaborative, in the browser, like Google Docs. Then we had a lucky stroke of genius — we showed those tools to our customers, and they immediately recognized the value, not just for maintaining web maps for shoppers but for leasing maps and security maps. Today tens of thousands of users maintain a live source of truth. Sign up, upload a floor plan or take a picture of one, the AI does most of the work of turning it into a useful digital map, and you author it from there — you don’t have to call somebody or outsource it to a digital agency.

Why Not Digital Twins?

Daniel: Some people hear “better than the PDF” and think: digital twin. You can carry a LiDAR scanner around and make a near-true copy. Why don’t you do that?

Hongwei: It’s an option — we have a free iOS app where, if you have a newer iPhone or iPad with LiDAR, you can create a realistic 3D capture of your space in a couple of minutes. It feels like magic. But what people don’t talk about with the digital twin narrative is: what problem are you trying to solve? If you’re driving down the highway using maps on your dash, are you using the map or street view? You’re probably using the map — top down, maybe a little 3D extrusion and landmarks. We humans require abstractions to function; I don’t want to see a LiDAR point cloud of the world, I want a map, because I just want to know where to go. And the map is so much easier to make and maintain than constantly recapturing. Digital twins are incredibly useful for selling property and virtual tours — it’s the digital successor to the architectural model the intern used to build — but nobody in their right mind operates a building based on a papier-mâché model. 90 to 99% of buildings in the world aren’t being built anymore, they’re operating on a shoestring budget, and they need maps that work and fit inside their current workflows.

Rethinking Wayfinding for Indoors

Daniel: Is the wayfinding user experience the same as Google Maps?

Hongwei: Google Maps for the indoors is still a good enough proxy. But a couple of things had to be rethought — and some of it is just hard, which is part of why this hasn’t happened yet. Outdoor mapping doesn’t really have the concept of multi-level — anyone who’s been to an airport knows GPS can’t tell whether you’re on the departures or arrivals level, which is confusing if your Uber pickup is on one or the other. The indoors is almost always multi-level. And most of the rendering SDKs everyone uses — Leaflet, MapLibre — weren’t built to render something multi-level in a lightweight way on a low-powered Android phone. We also had to incorporate landmark-based directions, because streets don’t exist indoors and north-south-east-west means nothing. We rethought a lot of these fundamentals — not because we planned to a priori, but because every customer makes us a little better; the Super Bowl and our airport customers, with some of the most dynamic spaces we’ve encountered, made us much better.

Firefighters Make the Most Maps

Daniel: How do the landmarks get in there — do you annotate them, or does the app pick them up?

Hongwei: We’re quite human-centric. The temptation is to imagine scanning everything and running computer vision on the plant, the fountain, on Santa — and you probably can, but it’ll still be at least 50% wrong, and that assumes you can convince the landlord to open every closet and let you scan everything, which is unlikely on private property. So we work with the data we already have: worst case the fire-escape map on the wall, average case a PDF someone already made once. We digitize it, OCR and computer-vision it, and give tools to the office manager, facility manager, marketing coordinator, and building owner — people who are doing this anyway. We launched a self-serve premium tool five months ago: you can draw maps by hand, or take a photo of a map hanging on the wall, and a computer vision model we’ve trained over three or four years vectorizes the raster image, cleans up the coffee stains and the architect’s signature, and annotates what it recognizes — doors, windows, room labels, fire-safety symbols, elevators, escalators, stairs.

Hongwei: This effort started in 2019, when we partnered with Homeland Security in the US to adapt our commercial tools for public safety. It turns out the firefighter in your city probably makes more indoor maps than anybody else, and nobody knows that. They have a binder in the back of the fire truck — plastic pages, floor plans with scribbles — because when they get called to a building they want some awareness in advance. That blew our minds. And we realized: if firefighters, who work out all day and run inside with an axe, can make maps, anybody can. For every professionally managed building there are eight map scribblers maintaining the same dataset — on paper. If we can get all those people collaborating on a digital source of truth, we get coverage. Firefighters making maps — that’s the mission, if there is one, at Mappedin.

Indoor Positioning and the Future

Daniel: Before there can be an explosion of indoor use cases, we need accurate indoor positioning. Where are we?

Hongwei: It’s hard — I cannot believe how many things have been tried that haven’t worked. Before Mappedin, my only real job was as an RF engineer at BlackBerry, testing Bluetooth and Wi-Fi circuits, and the main thing I remember is that this stuff is black magic. It’s basically impossible to use Bluetooth beacons to triangulate a signal — humidity changes it, your head is incredibly attenuating because we’re all salt water, one of the best blockers of RF signal. In the early days customers would ask “should I buy beacons?” and I’d keep saying no — and we watched customers blow about a million dollars each per building on beacons and then throw them away. Now it’s exciting to watch the giants — Apple, Cisco, Google — solve it at an operating-system level. Apple bought a company called Wi-Fi SLAM and uses passive Wi-Fi, accelerometers, compasses, gyroscopes, and ambient RF to build a statistical model of where the phone might be. It took a nation-state level of investment — the US government firing GPS satellites into space — to solve positioning outdoors, and we’re now seeing Apple, Google, Cisco, Samsung, Aruba invest comparable money to sell devices, with better indoor positioning as the exhaust. For a majority of our public-venue customers you can now have blue-dot positioning indoors.

Daniel: What does the future look like — when do we get to “one map everywhere”?

Hongwei: Twenty years ago people thought it would take 10% of the population working 10 years non-stop. Ten years ago we built tools that made the process about 10 times faster; last year we released Mappedin Premium, which made it another 10 times faster with automation and AI. So you could now map all of the indoors with 0.1% of the US population over 10 years — which still sounds close to impossible until you add up all the paper scribblers already doing it: the firefighters, building owners, office managers, insurance underwriters, city permit officers. With just 10% engagement from them, you could create the indoor base-map dataset over the next five years. I think in five years there will be cobbled-together base-map datasets, maintained by at least one person per building, and then you can build the ubiquitous applications that assume the map exists — the way you can assume today that a building has Wi-Fi. Public safety and insurance look like the use cases that drive 100% coverage. AR will be one of the applications built on top, but I’d be surprised if it’s a 10% use case anytime soon — indoor map adoption is about 20 years behind; we’ve got to get the ground floor and the foundations right before we can build the second storey.