News

Transforming Indian Agriculture in Andhra Pradesh

Some time before the Kharif season last year, the International Crops Research Institute for the Semi-Arid Tropics (ICRISAT) and the Andhra Pradesh Government decided to partner with Microsoft to bring some technology into predicting the best time to sow. Microsoft used its Azure cloud platform to obtain and bring together tonnes of data - going back some 40 years - about weather patterns in the area, soil conditions and more, built a modelling framework with ICRISAT, and applied its machine learning technologies. Microsoft also offered to send SMSes in Telugu to farmers on its findings, and ICRISAT informed the farmers that they had the option of going by those advisories. Farmers in Devanakonda mandal in Kurnool district of Andhra Pradesh are dependent on the rains. The area is not irrigated. The farmers are relatively poor, with landholdings of 2-3 acres each, and grow crops like groundnut and cotton.

Groundnut yields depend a lot on when it is sowed. It's best sown when there is good moisture in the soil. And when it starts flowering, the crop should not suffer a dry spell. Farmers go by their experience, gut feel, and weather reports to determine the sowing time.

Towards the end of May, the area received good rains, and some farmers decided to sow. The ICRISAT-Microsoft model found that it would be best to sow in the third or fourth week of June. About 175 farmers went by this advice. As it turned out, there was a dry spell in August, and those who sowed early suffered because their plants had flowered by then.

"Those who went by our advice saw an average 30 percent  higher yield than those who sowed early," says Suhas P Wani, director of the Asia Research Program at ICRISAT.

Encouraged by the experience, the programme has been expanded to 13 districts of Andhra Pradesh, and to many more crops, including cotton, ragi, rice, and maize. ICRISAT and Microsoft have also just entered into an understanding with Karnataka to develop an agricultural commodity price forecasting model, with the objective of enabling farmers to better predict prices. For this season, the prediction model will be applied to the tur crop.

"This year we should be touching 2,500 farmers in AP, and 1,200 in Karnataka. By next year, the number should be 10,000 in each state," says AVR Kesava Rao, honorary fellow and scientist (agroclimatology) at ICRISAT. ICRISAT and Microsoft are also building advisories for fertilizer and pesticide use, and for harvest and post-harvest practices, for each region, depending on its specificities.

Cloud computing, the enormous computing power that it brings, machine learning and artificial intelligence technologies, and extraordinarily powerful communication and collaboration tools - all of which today come at very low costs - are enabling societies to gather and analyze large volumes of data in a way that was impossible before, and deliver critical and useful information to even the least privileged.

Rajan Anandan, MD of Google India, says it's a matter of time before someone does in healthcare, education and agriculture what Airtel did in telecom - revolutionize the sector by bringing quality service at unbelievably low prices.
A senior Cisco executive notes that two-thirds of our people are in rural areas, but two thirds of our resources are in urban areas. "New technologies are enabling the transfer of the latter strengths to the former," he says.

Cisco was one of the early ones to attempt socially transformative projects in India using modern technology. In 2010, when floods ravaged parts of north Karnataka, the networking giant adopted five villages in Raichur district, rebuilt over 3,000 homes, and established four networked schools and a new digital healthcare centre. Teachers in Bengaluru delivered live video courses over the internet in English, mathematics, science, and social science to over 1,000 children in the newly-equipped schools. The healthcare centre enabled live remote consultations with good doctors in Bengaluru.

More recently, it won contracts from the Rajasthan Government to provide powerful communication and collaboration technologies to nine ITIs and to create 44 digital classrooms in schools and colleges. The purpose is to enable the students to access high quality content, and collaborate with quality teachers in urban areas. Cisco has set up a digital facility in Dharavi in Mumbai where artisans and leather goods workers can go to find global marketplaces for their products.

Anil Bhansali, MD of Microsoft India R&D, says the company has worked with the Andhra Pradesh government to develop a model that can now pretty accurately predict which student is likely to drop out of school, so that the school and public authorities can make appropriate interventions.

Similarly, Nandan and Rohini Nilekani's EkStep is using technology tools to measure the learning and performance levels of government school students, and determine the appropriate interventions required at the individual level. EkStep estimates it might have touched 1 to 3 million children through its partner organizations. Shankar Maruwada, co-founder and CEO of EkStep, says technology by itself may not solve issues, but it can significantly improve the capabilities of people, organizations, and governments involved in a sector to solve the problems.

A number of startups are working on a variety of models that exploit new technologies. IIT-Madras graduate Vivek Rajkumar's Aibono uses drones to do multispectral aerial imaging of plant leaves to understand the health of a crop, and then advises farmers on what NPK (nitrogen, phosphorus, potassium) fertilizer ratio to use. "A simple change in NPK ratio can alter production by a good margin," says Rajkumar, who has been funded by early Facebook investors Venky Harinarayan and Anand Rajaraman, and who works with more than 250 farmers in the Nilgiris region. As he accumulates more data, he is able to improve his prediction capabilities.

Flybird Agri Innovations uses sensors in the soil to detect moisture content and control irrigation, a venture supported by Chennai-based Villgro, a social impact fund backed by the Dell Foundation.

Rohtash Mal, a former CEO of tractor firm Escorts, has established what may be called an Uber for tractors. "Only some 10% of the country's farmers can afford farm machinery. How to reach the country's 90% was our question," says Mal. So his EM3 provides farm machinery & vehicles on demand, and farmers pay only for period of use. Technology allows him to ensure speedy service, and transparent billing.

Geetha Manjunath's Niramai uses thermal images and artificial intelligence to quickly assess the energy level in cells to conclude the existence or stage of breast cancer. And it can be done remotely and at a much lower cost than current diagnostic techniques. Qure.AI's algorithms can simultaneously look for and diagnose more than 15 diseases. Bengaluru-based SigTuple uses a small camera to read pathology samples kept under a microscope to analyze blood, urine, or semen samples. The data is then sent to a cloud server that can analyze and produce a digital report instantly. Great for places where doctors are difficult to access.

However, large-scale commercial models are yet to emerge. Cisco's early projects were CSR (corporate social responsibility) ones. Its then chief globalization officer, Wim Elfrink, was proud of the accomplishments in places like Raichur in Karnataka and Chhindwara in Madhya Pradesh. But, as per their agreements, when Cisco moved out of the projects and left them to the local authorities to run, the initiatives fizzled out.

Long-term success may require users to pay for what they use, and the participation of private entities who have the knowledge and incentive to run such projects. The Cisco executive says the country is at the phase of getting people to understand what technology can do for them, and large, successful projects will emerge in time. Microsoft's Bhansali says the company will move to commercial models once they affirm the basis of sustained success.



Share your comments