'BHU-MANACHITRA: NIT Rourkela researchers bag patent for drone system that generates real-time land maps without internet

Bhubaneswar: Researchers at the National Institute of Technology, Rourkela, have received a patent for an autonomous drone-based system that can generate real-time land maps without internet support or manual intervention. The technology, named BHU-MANACHITRA, combines Artificial Intelligence and Unmanned Aerial Vehicle capabilities.

Land mapping is crucial for identifying farmlands, forests, urban stretches and vegetation. It forms the foundation for infrastructure planning, natural resource management and environmental monitoring. In India, such work is still largely dependent on manual surveys that often take weeks or even months to complete. Although drones are being used to capture aerial images, the data usually needs lab-based processing, which delays map generation.

Sambit Bakshi, Associate Professor in the Department of Computer Science and Engineering, said existing deep-learning models often face difficulty reading complex terrains where buildings, roads and vegetation overlap. This results in unstable or inaccurate maps when used in real time.

Also read: NIT Rourkela’s indigenous, cost-effective device to help tackle heel pain.

To overcome these challenges, the NIT team has created a deep-learning model that enables the drone to identify land features instantly while flying. BHU-MANACHITRA performs on-board processing, making it useful in remote regions, disaster-hit zones and areas without communication networks.

Use of BHU-MANACHITRA

The technology can be applied by government agencies for land record modernisation, smart city planning and urban development. Agriculture departments can use it to assess crop conditions, soil health and irrigation requirements more quickly, improving efficiency in the field.

The system is also expected to support the Digital India Land Records Modernisation Programme, which aims to digitise cadastral maps across the country. During disasters such as floods, landslides or earthquakes, the drone can offer immediate terrain insights, helping authorities respond faster. Forest and environmental departments can use it to keep track of deforestation, encroachment and biodiversity changes.

Bakshi said the lightweight AI model powering BHU-MANACHITRA uses only 2.48 million parameters. Since drones cannot carry heavy hardware for real-time image processing, this compact model allows them to execute land mapping tasks with a small onboard processor.

He added that the innovation can significantly strengthen decision-making, resource management and disaster preparedness. With applications across various sectors, BHU-MANACHITRA has the potential to enhance national infrastructure and support quick, informed actions on the ground.