Github: apoorva-a98/IndoorDirection
Link to publication Link to portfolio website
As people have an inherent ability to navigate, assistance is only required to accurate the minor error in judgment of distance and direction (29 degrees) most minimally and efficiently. Hence, at any point, the information known by the user (full/part of the route, landmark, or even minor turn) should be given higher priority over the computed information.
Consequently, at every stage of a commute (before starting, while traveling and at the end) any parameter (knowledge of the route, mode of commute, landmarks on the way, intended destination, where to go next) affects routing differently and specifically. Hence the first principle, Any change in intent or knowledge by the user (about any part of the travel) at any point of the travel should be accommodated by the platform while routing.
The optimisation of routes seems to apply in three levels; 1–the quantifiable parameters (time, distance, direction). 2–the characteristics of the place (road-highway, residential). 3–potentially could be intent, work, chores (wife calling soon after office to get something), mood, etc. The second principle, Optimisation of the route at all levels – (1) the existing platform aided route (carrying quantitative information), (2) the physical road connectivity and characteristics and, (3) users’ needs’, should be given equal importance.
With the subjectivity with context and intent, time is the governing factor to crosscheck the feasibility of the instructions by the user and by the platform. The third principle, All decisions made by the platform (picking a busy route during office hours) or the user (going to Chickpet, Bangalore during rush hours) should be conforming to availability and suitability of time.
In today’s practice, the engagement of the user to the information provided by the platform doesn’t initiate nor supplement the experience ’off’ the platform. Fourth principle, The platform should not hinder or restrict the engagement of the user to the travel/journey.
Being confident about the two places the user has to travel between (start-end point, sub-point, etc.), a user, only when unable to achieve that (travel) themselves would depend on instructions by aid to help them achieve it. And the most important information (like, traffic will resolve in 20 min, there’s a grocery shop nearby) out of the provided information by the platform (traffic, intersections, halt time) is the one which helps the identified context (stuck in unmoving traffic). Finally, on having a lack of references, self-known and by the platform, we do depend upon the people around. A route can be formulated on tracing the location of these people-to-people interactions and guidance, leading up to the intended destination Finally, the fifth principle, A travel/journey should be approached as a set of sub-journeys, which begins and end with landmarks that act as – resemblance, place of interest or aid for the travel.
Navigation is the culmination of, ‘survey knowledge’—orienting ourselves to the destination—and ‘route knowledge’—familiarizing ourselves with and traveling a familiar path. Numerous outdoor navigation tools, such as AccessMap, Accessible Campus projects, Stockholm e-adapt, TP3, etc. are built over geoservices like Google Geolocation that use cellular and WiFi access points for accurate positioning. However, these methods falter indoors, where closed WiFi networks and static cellular access points hinder location detection.
Local Indoor navigation tools like BeMyEyes and SeeingAI focus primarily on immediate surroundings but lack solutions for ‘global indoor navigation’ (e.g., locating an elevator). Prior efforts for accessible indoor navigation either require high infrastructure setup(BLE beacons) or require navigating to sensors(NFC beacons, QR codes, and Tactile Signages).
In collaboration with NYU’s Ability Project and experts with lived experience, a major gap was identified: the need for accessible information on key landmarks (e.g., emergency exits, and bathrooms). Knowing these primary landmarks enables users to navigate secondary locations via local navigation methods.
This universal design problem highlights the need for tools that bridge immediate surroundings and broader spaces. We know users need error correction in direction and distance and, ‘survey knowledge’—familiarizing with routes by recognizing landmarks to realign themselves to their destination.
-
Plural Navigation Framework
humans have 30 degree error in direction
Various ways humans navigate the immediate surroundings - cane, wheelchair, sight, smell, touch - tactile signages, ai-bemyeyes
Various ways people navigate outdoors - signages,route planning, compass, turn my turn navigation, gps
Reverse Engineer
"There were many people who just gave a one liner saying the person does not have the
required skills."
"...You cannot automate networking..."
"...IBM complicates things, there's got to be a better way"
MacAddress |
Location |
6c:8b:d3:xx:85:xx |
north-side open studio |
dc:8c:37:xx:12:xx |
classrooms corridor |
dc:8c:37:xx:97:xx |
room 410 |
dc:8c:37:xx:f9:xx |
room 411 |
dc:8c:37:xx:83:xx |
room 412 |
dc:8c:37:xx:a1:xx |
room 413 |
6c:8b:d3:xx:84:xx |
entarace |
dc:8c:37:xx:b2:xx |
hallway behind the shop |
dc:8c:37:xx:e1:xx |
hallway by emergency exit |
dc:8c:37:xx:9c:xx |
south-side hallway |
dc:8c:37:xx:45:xx |
mid-floor open studio |
dc:8c:37:xx:73:xx |
south-side open studio |
Large indoor spaces like campuses, museums, and airports have closed networks, but as users connect, they can be detected by all routers (BSSID/Mac Address) in the area. AleNav, is a mobile application, that leverages this capability for accessible navigation. The application provides general area detection by tagging landmarks to the nearest router. Using voice, audio, and haptic feedback, AleNav vibrates when pointed in the direction of a landmark, then speaks out the landmark and distance. Users can also type or speak a landmark and scan their phone to orient themselves when it vibrates. The app announces nearby landmarks, aiding in building ‘survey knowledge.’ Orientation and course correction occur when the BSSID/router changes.
The project was developed by creating a dataset of BSSIDs from all routers at NYU ITP, tagging these IDs to landmarks. The application, built for Android, uses precise location to access BSSID data.
1. Arduino
2. MQTT (mosquito.net, public.shiftr.io)
3. Node.js
4. JavaScript
5. Android
Fix what you have and states
"There were many people who just gave a one liner saying the person does not have the
required skills."
"...You cannot automate networking..."
"...IBM complicates things, there's got to be a better way"
Fix what you have and states
"There were many people who just gave a one liner saying the person does not have the
required skills."
"...You cannot automate networking..."
"...IBM complicates things, there's got to be a better way"
Issues
Fixes
When the device connects to any router, there is a sudden surge of signal strength followed by a drop. After a few seconds, the signal strength relatively normalizes.
When the event listener is set to any state change, the message interval time here is a few milliseconds. This overloads my server during prototyping.
Since there is a constant fluctuation of +/-2 signal strength, I don’t really need to see that minute of a state change. I added a threshold of +/-5 to update the state change.
This helped with the surge issue on connection or switch to a new router.
Some routers have a higher bandwidth that the others which means you stay connected to those routers even when you move out of the location and closer to another router’s location.
Some routers are very difficult to connect to. This could be because of a another nearby router with a higher bandwidth due to which this routers has gone into hibernation.
Using WiFi routers (BSSID) for navigation could raise privacy concerns about tracking and data collection.
These can be mitigated by obtaining user consent to a privacy policy, implementing strict data anonymization protocols, and ensuring that no data is stored, only used for real-time navigation.
Fix what you have and states
"There were many people who just gave a one liner saying the person does not have the
required skills."
"...You cannot automate networking..."
"...IBM complicates things, there's got to be a better way"