“No one wants to intentionally hit an elephant,” says Seema Lokhandwala, an elephant biologist at the Elephant Acoustics Project, which focuses on developing technology to better understand the animal and mitigate human-animal conflict. And yet, train-elephant collisions are among India’s leading causes of unnatural elephant deaths, with the most recent incident taking place less than a month ago when a tusker lost his life in one in Assam’s Morigaon district.
“As far as official figures are concerned, nearly 200 elephants were killed in train collisions between 2010 and 2020, averaging about 20 elephants each year,” says activist and wildlife conservationist Neellohit Banerjee of Wildlife SOS, a conservation non-profit which is currently helping rehabilitate Bani, an elephant calf who was badly injured in a train collision last year.
Collision alert
In recent years, technology, particularly AI (Artificial Intelligence), has emerged as a game-changer in averting such accidents. Several projects are already underway. Last December, the Indian Railways announced the deployment of the AI-based Gajraj system along 700 km across multiple states. Earlier this year, the Tamil Nadu Government launched an AI and ML- (Machine Learning) enabled surveillance system along two lines that pass through Madukkarai in Coimbatore.
Early warning
Supriya Sahu, Additional Chief Secretary of Health & Family Welfare, Government of Tamil Nadu — who, in her earlier role in the Department of Environment, Climate Change, and Forests, was instrumental in setting up this system — says that since the tracks are located in the ghats where loco drivers cannot brake abruptly when they spot an elephant, a robust early warning system is essential. The new system consists of 12 high towers fitted with thermal and normal cameras, giving 150 m coverage on either side of the track.
Supriya Sahu inspecting a railway track at Madukkarai, near Coimbatore | Photo Credit: SPL
When an elephant approaches the tracks, an alert automatically goes to key stakeholders, including a patrol team, the forest department and designated officers in the railway department. “The railways inform the guard of the oncoming train, and measures are initiated: either reducing the train’s speed, stopping, or sounding the horn,” says Srinivas R. Reddy, Tamil Nadu’s Principal Chief Conservator of Forests and Chief Wildlife Warden. Also, since teams physically patrol these sensitive areas 24/7, they will try to get the elephant off the tracks and back into the forest.
Image captured by TrailGuard AI | Photo Credit: West Bengal Forest Dept
“Reaction time is really critical along a railway track. The faster you are to generate this information, the easier it is for them to react,” says Piyush Yadav, founder of Nightjar Technologies, a social impact enterprise that has developed and manufactures TrailGuard AI, an end-to-end, camera-based alert system.
Piyush Yadav
Forest officials say that since the system was installed in February 2024, over 500 animals, including elephants, have crossed the track safely. “For now, we are fine-tuning the system,” says Reddy, pointing out that AI and ML systems take time to offer tangible outcomes since they are dependent on the number of data sets they receive. “What we have found, however, is that this technology will be needed at other hot spots where we have human-animal conflict,” he says, listing out other locations in Tamil Nadu such as Kodaikanal, Pollachi, Gudalur and Hosur, where they plan to deploy this system.
‘There is no one concrete solution’
Other experts, however, are more cautious in their optimism, believing AI to be a useful tool but not a universal solution, given the complexity of the problem, one exacerbated by climate change and unmitigated development.
“Technology could play a significant part in facilitating solutions, but we really need to look at the root cause of why this is happening,” says Sandeep Kumar Tiwari, vice president and chief of conservation at Wildlife Trust of India, who has been working with various state governments, including those in Tamil Nadu, Assam Odisha, Kerala, to manage the issue. “It is a combination of various physical, technical and behavioural factors that animals are getting knocked down by moving trains. There is no concrete solution; we need to work on long- and short-term interventions.”
You cannot install one device and imagine all problems will be solved. Other factors, such as cost, scalability and the robustness of technology, must also be considered before installation. Moreover, one also needs to take into account the biology of an elephant — its behaviour and ecology — to deal with the problem. But some practical measures that can be taken, Lokhandwala says, include avoiding littering the tracks, moderating train speed, especially around hilly areas and curves and ensuring that any solution is customised to the local landscape.
AI in conservation
Understand the local context
M. Ananda Kumar, a wildlife biologist with the Nature Conservation Foundation (NCF), delves deeper into yet another factor that is often ignored in our conversation about human-elephant conflict: the local context — the landscape, how an elephant negotiates it and the communities that share space and resources with these animals.
Kumar, who has studied elephant behaviour and human-elephant conflict for over two decades, believes we need to understand the problem in the various places where the problem is acute before we develop context-based solutions. Since no two places are alike in terms of pressure on the elephants, their behaviour, land use pattern and so on, solutions need to be crafted accordingly. “People think that if you use technology in one place, it has to be applicable to all other landscapes,” he says. “But unless there is a strong scientific basis that supports replicability of technological interventions at the local level, they may not be effective in achieving desired results.”
preeti.zachariah@thehindu.co.in
Published - August 09, 2024 12:40 pm IST
