JWTE - Sudo Bash Bros - PathGuard

An AI-powered sidewalk assistant. With Proximity-sensing crossing lights, YOLO model for local traffic levels, and a local hazard map for rapid repairs.

Comment

Inspiration

Road safety has become an increasing concern for Calgary's pedestrians. As the frequency of collisions involving pedestrians increases, so does the demand for a solution. With Calgary scoring a 98% Walkability Score, innovative solutions like PathGuard are a requirement in upholding Calgary's high Quality of Life.

What it does

PathGuard has 3 main roles:

  1. On a local level, proximity sensors will turn on crossing lights when a pedestrian intends to cross an intersection such that, in a scenario where the pedestrian does not press the button, the lights will still turn on to alert passing vehicles.
  2. As a crowdsource initiative, users can report local hazards at each pedestrian crossing to alert schoolboards, community centers, or the city of Calgary, depending on the location, to alert repair teams for quicker response times.
  3. For easier data collection, a YOLO model will detect the number of people around community and park centers to help the City of Calgary in further urban development as well as allow residents to plan in accordance to live traffic estimates.

How we built it

Our team was split into the 3 standard tasks. Hardware ran on C++ in an Arduino environment. Roles 1 & 2 ran off an Arduino R4 w/ WiFi and Role 3 ran off a Freenove ESP32 WROVER. Backend used an AWS API Gateway and one of two lambda functions to store information in DynamoDB and trigger alerts via AWS SNS. The second lambda function connects with React.js & JSON for PathGuard's website, completing the frontend.

Challenges we ran into

With the hardware, one of the biggest challenges was using integrating two separate microcontrollers into a unified workflow. Because the ESP32's camera module occupied a majority of the GPIO pins, the remaining peripherals were required to operate off the Arduino R4, meaning separate main functions, control loops, and API endpoints. Additionally, the team stepped way out of its comfort zone by learning how to use AWS, DynamoDB, and React.js for the first time by applying it to our hackathon.

Accomplishments that we're proud of

Our project was quite ambitious, given the complexity of both the hardware and software workflows, letalone the challenge of integrating the two together. We're really proud of the fact that, by the end of the development period, we were able to complete all that we had expected to complete.

What we learned

Each team member had unique learning experiences. Alongside learning the various skills mentioned in the challenges section, we also learnt new tools like using PlatformIO to work on Git-powered projects and using WiFi protocols + YOLO resources on the ESP32.

What's next for PathGuard

Key tasks for the future of PathGuard include making a outdoor-resistant enclosure, custom hardware including a single control board, a second set of peripheral for the other side of the sidewalk, and community adoption.

Share this project:

Updates