B'luru techie’s dashcam spots potholes, complains

A Bengaluru engineer has developed a system using a car dashcam, GPS and AI to report potholes that are common on city roads. The technology detects damaged areas, records their location and size, and even creates a complaint in seconds. By connecting potholes to relevant road contracts, the system aims to identify the responsible authority for repairs.

One day of commuting led to the creation of this AI experiment for engineer Gaurav Sen. He fitted his vehicle with a dashcam, GPS and accelerometer. The dashcam captured the road and location data, while accelerometer tracked vehicle movement. Sen then used artificial intelligence (AI) to process the recorded information and send it to AI vision models.

The resulting system can analyze footage, identify potholes and classify them based on size - small, medium or large. Sen shared his project on Instagram on August 5, describing it as his method of reporting potholes to the Brihanmumbai Municipal Corporation (BBMP), the city's local government body. He emphasized the dangers posed by damaged roads, citing that over 2,000 Indians die each year in accidents caused by potholes.

The app performs more than just pothole detection. While an image of a damaged road provides visual evidence of the issue, a formal complaint requires additional details like the exact location and the responsible authority or contractor. Sen's system attempts to address these requirements.

The app searches through around 2,900 government road contracts to link a detected pothole with the relevant road project. It identifies the tender number, the contractor associated with the work and the government office responsible for the road. This feature could be particularly beneficial for roads still under warranty, as it may make the contractor responsible for repairs rather than treating potholes as new government expenses.

Preparation of a complaint takes only about four seconds. The system compiles the pothole image, exact location, tender details and relevant government office into a complaint record. Sen claims the process moves from simply spotting a pothole to generating the necessary documentation for reporting it without requiring manual input from the driver.

During his regular commute, Sen tested the system and it detected 12 potholes along his route. Each detection could be associated with photographic and location data, creating a record of the road damage rather than relying solely on a driver's description. The system also tries to link the road damage to the contract and authority responsible for that stretch, as identifying a pothole and getting it repaired are two distinct problems.

A civic complaint may not be effective if the reporter is unaware of the relevant agency or contractor.

Sen's project has gained attention due to its application of AI to address a daily issue faced by Bengaluru motorists. While the hardware used - a dashcam, GPS and accelerometer - are relatively simple, the AI component is what sets the system apart. The more intriguing aspect is the system's attempt to connect pothole damage to government contracts. This creates a chain linking the pothole, photograph, location, road contract, contractor and responsible office.

The project demonstrates that an ordinary commute can serve as a source of documented civic information. From detecting potholes to assigning responsibility, Sen's experiment aims to streamline the reporting process. The idea also opens up possibilities for cities to collect road damage data at scale, providing a more detailed picture of deteriorating roads and their associated projects or contractors.

For now, Sen's project remains an individual experiment, but the basic concept is straightforward: spot the pothole, record evidence, identify the responsible party and prepare the complaint for filing.