Inspiration
Imagine, you are in your first year of university. You are in a new city, new school, and new buildings. Google maps can only take you so far, directing you from building to building. But what about buildings that are far too big? How can one know how to navigate within a building to a specific classroom?
Far too often, students have difficulty navigating inside buildings they are not familiar with. Whether you are trying to find a new classroom, meeting up with a friend, or looking for your final exam room, we can all agree that this task can be cumbersome, anxiety-inducing, and overall unnecessary.
Thus, our team introduces MicroMapper, a way to micro-navigate the interiors of buildings to help first years and graduate students alike find the classroom they are looking for.
What it does
MicroMapper provides students an accessible way to find directions to their destination room within a building. A user inputs the building they are in, the destination room they desire to go to, and the room that is closest to them. MicroMapper then finds the shortest path from the inputted start to destination and displays it on a map. Now, there is no more need to guess if a room is in a particular direction only to take a walk of shame back.
How we built it
The React + Vite web application is hosted on AWS EC2.
Challenges we ran into
Trying to find a model that gives a good representation of a map for both input and output.
Accomplishments that we're proud of
Planning out the infrastructure of our application including frontend, backend, and database in a timely matter.
What we learned
How to use AWS EC2 to host a web application.
What's next for MicroMapper
Currently, our application is working for directions on the same floor level. Later, we want to add different floor level directions and accessibility options. Ideally, this application can also be extendable beyond UofT buildings to other campuses.
We will develop the backend for our application. We will use SageMaker to host our CNN model, which will be trained on a dataset consisting of floor plans and legends, as well as parseable grid formats. We will use Node.js for backend to process the shortest path and return directions to frontend which will display navigations to the end user. Additionally, we will allow for Administrators to input floor plans to the backend where the CNN will add grids to our floor plans for buildings.
We wish to utilize AWS S3 to store our building map data, where the CNN will store data and where our algorithm will retrieve data for a building.
Log in or sign up for Devpost to join the conversation.