Difficult roads are what led to the inception of the deep-tech start-up Roadmetrics

We have all been there — taken a shorter route suggested by Google Maps to either find a dead-end or a road that should come with its ‘Enter at your own risk’ sign. Then you curse yourself for trying to skip traffic on the main but at least ‘drive-able’ road and swear you will never again trust in unknown bylanes. But RoadMetrics believes that shouldn’t be the case. If there is a good alternate option, a commuter can now see for herself if the road is smooth (indicated by the colour green) or a nightmare (shown in red) on the app.

And the idea behind this solution came from two engineering graduates — Dipen Babariya and Mishal Jariwala — who took an alternate route and spent double the time to reach their destination, two years ago. They realised how Google Maps doesn’t do the trick when it comes to Indian roads, which are vast and in many cases, inefficient or pothole-ridden. So, when they moved from Gujarat to Bengaluru last year, they met US-return Nikhil Prasad Maroli, an industrial engineer who has worked with Velodyne LiDAR and Tesla. Thus, RoadMetrics — an app that detects all major road defects — was founded in October 2019 and raised seed fund of Rs.2.5 million from 100X.VC in December.

Using the power of machine learning and AI, the app automates road condition assessments and mapping. The process is fairly simple. “A smartphone, which has our app downloaded, is mounted on a car’s dashboard. As the car moves, an image is captured every 10 feet. This image is then sent to our server on the cloud,” says Maroli, while explaining that they have signed up with Amazon Web Services for its low-cost server infrastructure. Suppose there is a speed-breaker without markings on it. This will obviously make image mapping difficult. In that case, the platform uses sensor detection with the help of accelerometer and gyroscope — a tech that is installed in all smartphones today.

There are challenges, admits Maroli. “The mapping has to be done when there is good lighting and no traffic,” he says. The team has covered over 2,500 kilometers in the cities of Surat, Bengaluru and Jamshedpur. Its test project was in Surat with the city’s municipal corporation. In Jamshedpur, the start-up registered its first earning of Rs.25,000 when it tied up with Tata Group’s JUSCO for a pilot project. With a team of four, RoadMetrics is trying to automate the updates as much as possible. They also plan to implement LiDAR imaging in the software soon that will help with depth perception of potholes and give a 3D perspective of the roads. “Going forward, we are also looking at incentivising people who can provide us with crowd-sourced data using our app,” says Maroli.

With the 100,000 data points from three cities, RoadMetrics has been able to achieve an accuracy of 90%. But, with a little over 600 downloads on Play Store, RoadMetrics co-founders know that it is tough to take on Google Maps. Thus, their current focus is on the enterprise app for larger businesses. “With the enterprise platform, firms can download all the mapped data and use it for prioritising maintenance and allocating budgets,” says Maroli. For instance, a logistics company or a local municipal corporation could benefit from their data to make their work more efficient. Mapping companies could also use its API in their application to send alerts to drivers.

Besides a handful of municipal corporations, infra and logistics companies, the team is also in talks with Mahindra Group for projects in Jaipur and Chennai. They are aiming to map ten Tier-I cities by early 2021 and are estimating all the projects in the pipeline to bring revenue worth Rs.19.5 million for the current fiscal.