EcoTrack

Your commute, our commitment.

Comment

✨ Inspiration

We have had many painful experiences going through traffic jams on Swiss roads. In this hackathon, we wanted to think of a sustainable solution that could relieve the suffering of the driver and the passenger(s) or a way to reduce the congestion. Introducing: EcoTrack 🛣️

❓ What it does

Our app helps to motivate drivers to change their behavior by awarding them with scratch cards, coupons, and other benefits to pick a more sustainable route or slow down to reduce congestion. It has multiple features:

  • Traffic flow detector: Users can insert their starting point and destination, then our system will predict if there'll be a traffic jam ahead
  • Departure time recommender: Select the planned departure time, then we can suggest ones with less congestion and a route that is environmentally friendly :)
  • Prize inventory: Collect coupons, vouchers, and discounts by using the app and driving safely

💻 How we built it

  • We developed a neural network model with PyTorch and Scikit-Learn to predict traffic flow based on the traffic data from BIT
  • Tested different approaches (multi-layer perception models) and experimented with various error/loss functions to optimize our model
  • Deployed an online user interface (UI) using modern web and mobile development technologies such as Typescript, React, and Vercel
  • Filtered and interpolated the given datasets to achieve the best possible accuracy and minimize the error
  • Created an algorithm to calculate the best departure time

For more info on the models and the backend, go to this Github repo. If you're more interested in the web/mobile app or "the pretty stuff", go to this repository instead :D

⚔️ Challenges we ran into

  • Extracting and cleaning the huge amount of data was challenging
  • We didn't have a lot of computation power or the time to go for stronger models
  • No "obvious" solution - we had to be creative along the way

🏅 Accomplishments that we're proud of

  • We managed to get a few hours of sleep
  • Developed a novel framework to preprocess and clean the traffic data
  • Successfully deployed an online web & mobile platform (https://hack-zurich.vercel.app/) within less than 24 hours!
  • Racked our brains in search of possible gamification elements in encouraging sustainable commuting habits. Previous ideas include: memes, ghosts to scare bad drivers, and VR plants that grow along your trip
  • Achieved a 0.0000000049 mean squared error (MSE) loss on our own test dataset
  • Made a nice video for our pitch ;)

📚 What we learned

  • Split and assign the tasks early!
  • How to preprocess and analyze large-scale traffic data efficiently
  • Ways to deploy a working prototype & the importance of user-centric design
  • How to find the quietest spots to record the video

🔮 What's next for EcoTrack

Winning HackZurich 2023

🤗 Meet our team!

  • Viet Duc Kieu (Vietnam)
  • Yohan Thibault (France)
  • Sara Rutz (Switzerland)
  • Suchit Gupta (India)
  • Nathanya Queby (Indonesia)

🗣️ Fun fact

We all come from different countries and two of us took the highway to get to Zurich 🛣️

Built With

Share this project:

Updates