According to recent reports, former Indian Space Research Organisation (ISRO) scientist Madhukant Patel has developed an AI-based soil analyser. This AI-enabled device provides a report on soil health within seconds, instead of the current time of a fortnight or more.
The AI-based soil analyser
As reported by the Times of India, Patel recently presented the prototype to top officials of the agriculture department, which has asked all the soil testing laboratories, along with IFFCO and Gujarat State Fertilisers and Chemicals (GSFC), to provide the scientist with 5,000 soil testing reports so that the instrument can be calibrated.
"The device uses ultraviolet, visible, and infrared light to determine the presence and quantity of nutrients in the soil. With the help of nanotechnology, metal rods have been developed to measure other soil properties. The device can also measure organic matter, humus, and carbon content, which are crucial for natural farming," Patel said.
Also read: ‘We Are In Default Alive State With 1200 Crores In The Bank’: Unacademy CEO Gaurav Munjal (startuppedia.in)
How does the AI-enabled device work?
As claimed by him, the device can detect the presence of beneficial bacteria and fungi, benefiting soil preparation for natural farming.
Patel disclosed that the data captured by the device is gathered, and then artificial intelligence and machine learning tools convert the data into a spectrograph.
The handheld device works on the principle of spectroscopy, generating a soil health report card within seconds of placing its sensors in about 400 g of soil.
Traditional soil testing in government labs involves a ‘wet chemistry method' involving grinding, heating, and applying chemicals to soil samples.
On the other hand, the ex-ISRO scientist’s device allows on-site testing, eliminating the need to transport soil to labs.
Also read: Govt Shuts Down 75 Ola Electric Stores In Maharashtra Operating Without Trade Certificates (startuppedia.in)
/startuppedia/media/media_files/2025/04/17/pWyNdOvVlQVfsWpnBlyG.jpg)