This is my sketch for a direction-generating system for ADA-compliant buildings. The goal is to provide a simple interface to people with visual impairment that will provide them with human-comprehensible, non-metric (i.e., 'Go straight 4 doors, then turn left' instead of 'Proceed 23 meters, then turn west') directions to help them get to where they need to be.

It is (generally) divided into 3 parts:

SLAM : using Monocular SLAM, build a map of a space, with each image corresponding to a pose.

ImageComprehension : Go through images and pick out those that contain an ADA-compliant room marker, including stairs and exits. Generate an adjecency list, with each node as a spot in the hallway. Can have rooms, other nodes, or exits/stairwells to North, South, East, or West.

DirectionHandling : Given a start room and a finish room, generate human-comprehensible, non-metric directions. At first, the nearest beginning room must be provided.

Not ready for a release yet :^(

• John Feilmeier - Coding and research - Jacktavitt

See also the list of contributors who participated in this project.

This project is licensed under the MIT License - see the LICENSE.md file for details

• Billie Thompson - README template - PurpleBooth *Excellent template