InterNAV

Inspiration

Many people at HTN 2019 are university students from other parts of the world, often with no experience of the layout of the area and relegated to having to haphazardly navigate looking for their preferred destination. This trend continues after university, as these students, now employees, enter large mazes of company buildings.

While many applications and services exist for larger-scale traversal, no such service really exists when it comes to internal navigation; there is no Google Maps of the insides of a building. Thus, we were inspired to create InterNAV.

Objective

The purpose of InterNAV is to show a user's location on a map of a building's inner layout and present the shortest path towards a supplied destination.

Development Process and Methodology

For every corner in the building, we recorded what's called a "reference point", where we get the decibel strength of all the surrounding WiFi access points. Then, we take the decibel strengths of the access points near you, and use machine learning to find the location on the map that will minimize your error from a node. Then, we use the difference in error to interpolate your exact position. Since we have a modeled graph of the building using reference points and their connections, we find the shortest path by implementing Dijkstra's shortest path algorithm.

Challenges in Development

Our main challenge in developing InterNAV was figuring out how to make location detecting accurate. Many times, our location would wildly fluctuate within anywhere of 20 metres of our actual location. After some optimization, filtering extraneous data, and finding the most accurate ways to reduce errors, our accuracy was refined and our average error ranging up to a single metre.

Accomplishments and Lessons

Ultimately, our greatest accomplishment in the development of InterNAV was the incredibly streamlined and efficient end result we produced at the end of the time allotted. Our project ran much smoother and more intuitively than we had originally envisioned; the algorithms implemented worked as expected with minimal issue, after many hours of schematics and planning. The UI was rich with visual manipulation options such as scrolling, resizing and zooming with responsive controls, which took a lot of fumbling with our given tools. Against the numerous problems we faced, we were able to develop a complete, technically complex hack.

Next Steps/Future Plans

• Mobile application
• Expansion to entire building complexes
• Higher responsiveness through more optimization

Built With

• firebase
• locator
• machine-learning
• math
• pygame
• python