Navistride

Get paid to capture real-world driving data, with your phone or our own device. Data is then sold to train and improve autonomous vehicle systems with authentic, unsimulated road conditions.

Comment

Inspiration

I've always wanted a self-driving car here in the Philippines, but most AV companies only release their cars in wealthy cities around the world. I didn't think we'd ever get self-driving cars here, given all the issues on our roads, like non-functional traffic lights, people walking in the middle of the street, potholes, and more. It seems the Philippines isn't ready for the self-driving future.

But apparently, even in countries where AVs are already deployed, the AI still isn't ready. These cars continue to struggle with reliability, making them far from viable. Compared to the rapid advancement of LLMs and AI art, it's clear that self-driving AI is missing something in its training: the bad data, the edge cases, the unpredictability of human environments.

That’s why I created Navistride with the goal of collecting real-world driving data that can help autonomous vehicle companies train systems that truly understand the chaos of reality, and advance self-driving cars at the pace of modern AI.

What it does

Navistride enables everyday drivers to passively collect real-world driving data through a mobile app (or future hardware device). Just put the phone on your dashboard, turn on the internet and GPS, and record, you'll get paid for the time, distance you've travel, and other external factors. No changes to your behavior. No specialized routes. Just your everyday driving.

Data received will then be used by self-driving companies to train their AI. Think of it as Outlier or Appen but for self-driving cars instead of generative AI.

How we built it

The website and demo app is completely made using bolt.new using React.js, Vite, Typescript, Tailwind CSS, Framer motion, and Supabase. 90% of the code is made using AI, including the working Camera+GPS recording and map tracking, as well as the backend use of Supabase for waitlist registering, demo login, and user earning tracking. 10% of the human intervention are for the logos, videos and images and minor text changes and bug fixes.

Challenges we ran into

An actual hardware isn't really gonna be built with just a week before the deadline, so I thought of a mobile camera and GPS compromise, and luckily bolt.new's AI was able to help build it on the web. There are still some performance issues with other browsers and in older mobile phones, but the app should work in most modern phones

Accomplishments that we're proud of

This is a type of project that would have taken months to finish... I made Navistride with bolt.new in three days.

What we learned

AI can now build sophisticated, full-stack applications as well as creating professionally designed websites using framer.

What's next for Navistride

If I win this hackathon, my first priority will be building the prototype 360° camera and LiDAR device myself. I believe deeply in this project and in its potential to provide the kind of real-world data self-driving AI needs to reach a truly autonomous future. I also plan to onboard a few real users from the waitlist and use part of the prize money to pay them for their data, proving that the system can work, even at a small scale.

Built With

Share this project:

Updates