Skip to content
A mobile based product which records the road quality in runtime and processes on algorithms indigenously. App made for HackInOut 2018.
Branch: final_product
Clone or download
Type Name Latest commit message Commit time
Failed to load latest commit information.
.gitignore added logs and made model to load once Oct 8, 2018

What is the problem?

The idea originated once when my(Kaustubh) grandma was travelling from our home to a place around 65 kms. Though, we knew many paths to reach the same place (from GoogleMaps), but the problem in front was not the path or the distance but rather the quality of road as whole (one or two potholes didn’t matter but rather the longer road patches of good / bad roads). This problem is daily faced by patients and old people especially with problems like slip disc.

What do we propose?

We propose to build an mobile based product which records the road quality in runtime and processes on algorithms indigenously built by our team.

Though there are many research papers and products around in the market which are working in same direction but :

  • Most of them are taking data using dedicated attachment to vehicles like car which is expensive as well as infeasible to be installed in such large number of automobiles
  • Above 90% are developing algorithms of “Pothole Detection” are for the foreign countries where there are hardly any potholes on the road.

But in countries like India with roads in such condition, every point would be detected as a pothole with these algorithms, and it is infeasible to install dedicated devices on every vehicle.


Our first aim is to provide assistance service to hospitals and ambulances. Whenever an ambulance would be used, the app will take input of patient's condition and display the most preferable path. For say :

-> If a person has suffered an heart attack then short routes would be preferred above smoother ones.

-> For a person who suffered some injury for say a bone-breakage a smoother route will be preferred.


We have spent most hours developing algorithms that give results for real life situation. Going through around 17 research papers did gave some ideas but none of them have been helpful for real life conditions.

We have come up with an indigenous algorithm for dividing road patches in 3 types :-

  • Village roads (10-20 km/hr)

  • City roads (30-40 km/hr)

  • Highways (60-70 km/hr)

Our current app is working perfectly fine for detecting all road conditions though we couldn’t give time for UI development or development of our final product.

Also as a part of safety service we have also implemented an ML model which will alarm you of upcoming big potholes when you are speeding high.


In future we plan to extend our app for all users to improve the driver experience.

Download the apk here :

You can’t perform that action at this time.