Bengaluru-based software engineer and entrepreneur Gaurav Sen has demonstrated an artificial intelligence-powered system that can detect potholes from a moving vehicle and connect road defects with government contracts, contractors and officials responsible for the concerned stretch.
Sen, founder of technology education platform InterviewReady, developed the system using a dashcam, GPS and accelerometer data. The setup records road conditions while a vehicle is moving, after which AI-based computer vision is used to identify potholes and assess their size.
The system is also designed to distinguish potholes from other road features, including speed breakers, reducing the need for users to manually identify and document individual defects.
Its functionality extends beyond pothole detection. According to Sen's demonstration, the software can search through roughly 2,900 government contracts to identify the tender associated with a road, the contractor responsible for the work and the relevant government officer.
The system then combines this information with a photograph of the detected pothole and its geographical coordinates to prepare a structured civic complaint.
During one drive shown in the demonstration, the system detected 12 potholes and generated 12 complaints that were ready to be filed. The demonstration, however, does not establish how accurately the system performs across different roads, weather conditions or larger datasets.
The project adds an accountability layer to a category of technology that has traditionally focused on identifying and mapping road damage. By connecting visual detection with government contract data, the system attempts to link a physical infrastructure problem to the administrative information associated with the road.
Bengaluru already has digital mechanisms for reporting road infrastructure problems. The former Bruhat Bengaluru Mahanagara Palike's Fix My Street system allowed residents to submit geo-tagged photographs of potholes and route complaints to officials. The Greater Bengaluru Authority also lists a Fix Pothole service among its civic applications.
Sen's system differs by attempting to automate several steps between identifying a pothole and preparing a complaint, including capturing its location and searching contract records for information about the parties responsible for the road.
The project comes as computer vision and generative AI are increasingly being tested beyond conventional chatbot and content-generation applications. Image-recognition systems can analyse visual data from cameras, while AI tools can process larger datasets and combine information from different sources.
For civic infrastructure, similar technologies can potentially support road-condition monitoring by processing footage collected from vehicles and converting it into structured information. The effectiveness of such systems, however, depends on factors including detection accuracy, access to updated government records and the quality of underlying contract data.
Sen's demonstration has attracted attention online, with discussions around whether the approach could be adapted to other cities and civic issues.
For now, the project remains a demonstration of how AI, location data and public contract records can be combined to move pothole detection from identifying a road defect towards generating information that citizens could use when filing civic complaints.