We were recently offered the opportunity to provide indoor navigation for one of Hong Kong’s most anticipated annual conferences, the 2023 Hong Kong Fintech Week. Well over 10,000 participants were expected, and the conference’s hundreds of exhibitors would sprawl over more than 17,000 square meters.
We were, of course, tremendously excited to show off our technology to such an audience and in such an iconic location as the Hong Kong Conference and Exhibition Center, but undertaking this task came with a couple of tough constraints.
Ultimately, we overcame these challenges and provided a successful navigation experience, helping hundreds of people find their way and each other.
In this post, I want to highlight the challenges of indoor positioning, especially for events and conferences, and show how the posemesh was uniquely well suited to enable the experience.
Fundamentally, indoor positioning is difficult because digital devices have a very poor understanding of their place in the world.
Although it is theoretically possible to describe the precise GPS location of every individual booth at the conference, it is still practically pointless because the navigating devices themselves have such a limited understanding of their position relative to that coordinate.
1. Time Constraint: The conference organizer only gets access to the venue a day or two before opening, and that’s the theoretical maximum setup time available. Practically, however, the venue is not fully set up with booths and banners until the very last moments before opening, which can reduce the implementation window further.
2. Cost Constraint: Although Fintech Week is a big event, there are reasonable limitations to how much capital can be deployed to provide navigation for a few days. Any positioning system we use for the navigation has to be deployed both within the time and cost window afforded by the venue.
3. Data Constraint: Every major conference in the world is plagued by bad internet, which will manifest as two distinct constraints. One is the amount of data that can be reliably transmitted over a reasonable amount of time, and the other is the amount of time you have a reliable connection. The data constraint means that we need a positioning solution that minimizes the amount of data that has to be transferred in and out of the navigating device.
4. Connection Constraint: The total data is only part of the picture. We must also ensure the solution works in pockets of spatial or temporal internet shadow. The navigation solution has to work in as many cases as possible while dealing with potentially patchy connections.
There are several ways that one could attempt to provide the necessary positioning, but the most likely, common and practical approaches are:
Hardware Beacons: Physical devices scattered across the domain used for trilateration. Typically Bluetooth or UWB.
Wifi/Electromagnetic Fingerprinting: Machine learning models trained on the electromagnetic environment of the domain.
Visual Positioning: A pre-scanning of the environment creates a 3D reference that the device camera feed can be compared to.
Hardware beacons do well with the data and connection constraints but do less well with setup time and cost for impermanent installations. Dozens to hundreds of beacons would be required, which would have to be sourced, installed, and then taken down again. Hardware beacons are also not the best solution for device compatibility, and you might limit your reach.
Wifi/electromagnetic fingerprinting is a more novel approach that, in the best scenarios, gives a meter of accuracy (which is enough, realistically, for the use case). But before such a fingerprint model can be trained, you first have to capture the fingerprint - and the fingerprint is unlikely to remain constant in an environment that is physically changing so much before opening. Additionally, it was unclear how much impact there would be from the thousands and thousands of devices moving through the space together. The solution would have done well with data and connection constraints, just as the beacons would have, but the feasibility of training the model in time (or even having a reliable environment at all) was a risk factor.
Visual positioning would most likely be the default approach of most of our competitors but suffers the most under the constraints. It’s impossible to map the environment until the last hour before opening. Moreover, many visual positioning models may struggle to recognize the environment when it is densely crowded, and the solution is very reliant on good internet - so reliant, in fact, that we always (gleefully) ask competitors to demo their AR multiplayer solutions at AR trade shows because most of the time they cannot! The digital twin of the 17,000 sqm can realistically be on the order of hundreds of megabytes to a few gigabytes, and the reliance on sending camera-derived information to the VPS system makes it much too dependent on the internet connection.
We instead took a marker-based approach on the posemesh, placing 110 unique QR code “lighthouses” across the domain and mapped their spatial relationships to each other.
Essentially, our “map” of the event space was 110 unique named poses in a coordinate system and a navmesh of walkable areas expressed as quads.
The whole domain was represented in under 250kb of data compared to the hundreds to thousands of megabytes required for spatial anchors.
This means that we can overcome both data and connectivity constraints, making use of local caching and benefitting from the minimal data exchanges required.
The markers are also independent of the layout of the conference being finalized. Several markers could be placed straight away, but, in fairness - many codes were damaged during booth setup due to the chaotic nature of the setup process.
Like the other approaches, we had a minimal time window to implement, but we successfully placed the 110 QR codes and mapped their relationships well before the conference started.
Another benefit of the QR codes is that they help with discovery - the codes help redirect new participants to the relevant app experience and information they need to get started. Thousands of scans were recorded even on the first day, and hundreds were converted to app downloads with completed navigation sessions.
The posemesh also helped distribute traffic across its decentralized networking architecture, increasing the likelihood of each participant having a good experience.
To deliver an indoor navigation experience to a large conference, we turned to marker-based positioning on the posemesh.
110 unique QR codes were placed on the venue floor, driving thousands of scans and calibrations, hundreds of downloads, and successful navigation sessions.
The solution was implemented the day before opening, required only 250kb of mapping data, and the capital expenditure was the cost of printing 110 scratch and glare-resistant markers.
In summary, marker-based positioning makes it possible to deliver fine-grained indoor positioning in challenging and changing environments, even when there are strict constraints around time, cost, connection, and data limits.
Auki is building the posemesh, a decentralized machine perception network for the next 100 billion people, devices and AI on Earth and beyond. The posemesh is an external and collaborative sense of space that machines and AI can use to understand the physical world.
Our mission is to improve civilization’s intercognitive capacity; our ability to think, experience and solve problems together with each other and AI. The greatest way to extend human reach is to collaborate with others. We are building consciousness-expanding technology to reduce the friction of communication and bridge minds.
X | LinkedIn | Medium | YouTube | AukiLabs.com
The Posemesh is an open-source protocol that powers a decentralized, blockchain-based spatial computing network.
The Posemesh is designed for a future where spatial computing is both collaborative and privacy-preserving. It limits any organization's surveillance capabilities and encourages sovereign ownership of private maps of personal and public spaces.
The decentralization also offers a competitive advantage, especially in shared spatial computing sessions, AR for example, where low latency is crucial. The posemesh is the next step in the decentralization movement, responding as an antidote to the growing power of big tech.
The Posemesh has tasked Auki Labs with developing the software infrastructure of the posemesh.
X | Discord | LinkedIn | Medium | Updates | YouTube | Telegram | Whitepaper | DePIN Network
