Researchers get AI help to map ecosystem, wildlife conservation
In North East India’s biodiversity hotspots that are home to rare flora and fauna such as the Hoolock Gibbon and the Phayre’s Leaf Monkey, a Bengaluru-based research organisation has been gathering data on plant diversity and animal groups as well as mapping the ecosystem. Alongside, it has been wondering how to make sense of the data collected, given the limited resources at its disposal.
In Hyderabad, researchers have been trying to devise a method to predict a pest attack, so farmers can be warned in advance.
In both cases, artificial intelligence (AI) has come to the rescue. “Our researchers are developing a machine learning algorithm in the cloud space of Microsoft Azure to facilitate ecosystem mapping and to create a bioresource atlas for the region,” says Nitin Pandit, director of Ashoka Trust for Research in Ecology and the Environment, the organisation working in Sikkim.
He adds, “The shortage of trained human resources in the documentation of biodiversity is a major impediment and AI tools bolster the efforts of documenting biodiversity data.”
The application of AI in our daily lives—it’s the reason why every email lands in the right Gmail folder or Amazon recommends products suited to your taste—is well known. What’s new is that AI-based technologies are catching on big time across Indian research organisations working on environment and conservation. These organisations are being backed by big-ticket funding programs such as Microsoft’s AI for Earth, a $50 million, five-year program that has eight participants from India. Support has also been forthcoming from the central government, which recently approved a Rs 400-crore National Artificial Intelligence Centre that is scheduled to begin functioning from July this year.
Lucas Joppa, chief environmental officer of Microsoft, says, “AI is helping organisations understand life on Earth—where species are, how many there are and how their behaviour is changing over time as their habitats change.
This leads to more effective protection programs. India’s response is exciting with the third largest concentration of grantees for the AI for Earth program located here, just behind the United States and Canada.”
AI to prevent poaching
Till date, the most popular AI application on protecting wildlife has been PAWS or Protection Assistant for Wildlife Security. Designed by Fei Fang of Carnegie Mellon University in Pennsylvania, US, the AI-based tool forecasts wildlife poaching based on ranger-collected data and evaluation through field tests. “The program predicts a heat map of poaching threats. Over eight months of using PAWS in Uganda, it was confirmed that the predictions made led to forest rangers catching poachers or detecting snares and traps at a higher rate than before,” says Fei Fang.
“Every snare removed means a potential saving of elephant life. It’s not just about making predictions about poaching threats but PAWS also assists rangers to design the most effective patrol routes, even in tough terrains,” she adds.
This enables the most effective use of manpower available. PAWS has been successfully applied in South-East Asia as well as China but has not made it to India yet. However, companies such as Teradata India are working with the Indian government to apply AI to prevent poaching.
Souma Das, managing director, Teradata India, says, “We are actively engaged in wildlife conservation and protection using AI-driven predictive analysis on wildlife movement, tracking and monitoring their health to avoid the spread of any disease. We are also protecting wild animals from poachers, through continuously monitoring thermal and video images and analysing uncommon behaviour using machine learning techniques.”
Apart from preventing wildlife crime, AI is also being used to protect habitats. Chintan Sheth, a researcher at Bengaluru’s National Centre for Biological Sciences, has been using AI to predict habitat cover in the Eaglenest Wildlife Sanctuary in Arunachal Pradesh and Chilika Lake in Odisha.
“AI makes the job easier and faster, especially when used within the Google Earth Engine platform. I no longer have to look at every satellite image, just some thoughtful code can predict land cover change across 30 years. Google Earth Engine is a powerful cloud computing platform that uses thousands of computers on the network to run analyses, making it phenomenally fast and efficient,” says Sheth.
AI for farming
At Hyderabad’s International Crops Research Institute for the Semi-Arid Tropics (ICRISAT), researchers have been applying AI to forecast pest attacks on crops and advise farmers on the best times to sow. Pest attacks are estimated to cause an annual crop loss of up to $500 billion for Indian farmers.
Says Mamta Sharma of ICRISAT, “Our project is a combination of better surveillance powered by satellite images, better forecasts powered by machine learning, and a robust decision-support system for farmers through a cloud-enabled app. Advanced pest forecasting models and GIS maps are having a significant impact on the understanding of constraints in crop production and to assess the impacts of climate change in the near future.”
She adds, “To ensure smallholder farmers have access to information in time, we are also planning to link our pest prediction models with an AI-supported mobile application that displays personalised prediction results and recommended actions for each farmer.”
By using the AI sowing app, the research institute claims that farmers who took part in a pilot program in Andhra Pradesh had a 30% higher yield per hectare on average. Sharma says, “The app sends sowing advisories to farmers about the optimal date to sow. The farmers don’t need to instal any sensors in their fields or incur any capital expenditure for this. All they need is a feature phone capable of receiving text messages.”
IBM, too, has jumped on to the bandwagon and is aiming to reach out to as many as 3.5 million farmers through its AgriTech apps. “We see AgriTech as a Rs 5,000-crore opportunity in the next five years. We are also working with Niti Aayog to develop a crop-yield prediction model using AI to provide real-time advisories to farmers in backward districts,” says Himanshu Goyal, India business leader, the Weather Company, IBM. Having made large strides in the field of environmental sciences in India, AI is now a necessary tool. Joppa of Microsoft says, “In one of our projects, there are 50 sensors monitoring a 1,243-sq km area to listen to elephants as well as to more troubling sounds like gunshots or chainsaws. They record a lot of data, about two million songs and calls from the forests every 3-4 months.”
He adds, “With the help of AI, the data is being processed in a few weeks where it would have taken a couple of months earlier. That’s the kind of efficiency AI offers. No human would be able to sit there and listen to two million songs in a language they don’t understand.”