The death of seven elephants after the Batticaloa-Colombo train Meenagaya collided with a herd of elephants crossing the railway track from the Kaudulla National Park has led to renewed interest in solving the island’s longstanding saga of Human-Elephant Conflict (HEC).
Train collisions involving elephants are not new to Sri Lanka; just last year, a train carrying thousands of gallons of fuel hit a herd of wild elephants in Minneriya, renowned for its nature park and wildlife, killing two and injuring one other.
Minneriya National Park is part of the elephant corridor that connects the Kaudulla and Wasgamuwa National Parks, drawing thousands of tourists each year hoping to see elephants in their wild habitats.
Similarly, in 2023, six elephants were killed in train accidents in just one night in two separate train collisions – four died after colliding with a night mail train travelling from Colombo to Kankesanthurai, while a female tusker and a calf succumbed to injuries after colliding with the Meenagaya train en route from Batticaloa to Colombo.
There have been many attempts throughout the years to solve HEC and elephant deaths caused due to train collisions in particular, given their impact on conservation.
A 2024 study by researchers from the University of Colombo using data from 2010 to 2023 on identifying collision hotspots notes that most hotspots were located in areas where railroads traversed or were in proximity to protected areas, or areas heavily used by elephants. It points out that implementing targeted mitigation measures at the identified hotspots may help reduce the incidence of elephant-train collisions in the country.
Department of Wildlife Conservation (DWC) Director Operations Ranjan Marasinghe attributed the proliferation of such collisions to changes in elephant travelling patterns, likely due to climate change.
“Previously, collisions that occurred when elephants crossed from Minneriya to Kaudulla and vice versa were confined to a certain period,” he said, indicating that there appeared to be an increase in the frequency of their travel at present. “Currently, they appear to be travelling back and forth between these two locations daily.”
Pointing out that such accidents had averaged at around 11-15 per year historically, he noted that 2023 had witnessed a sharp increase in such deaths.
Stressing that the department was taking all possible measures to reduce such incidents, he added: “This is not something that the DWC can undertake alone. The department extends the necessary support for initiatives, such as widening the mounds through which trains travel, allowing elephants to move aside when trains pass.”
Similarly, General Manager (GM) of Railways J.I.D. Jayasundara told The Sunday Morning: “I believe the average number of elephant deaths due to train collisions is around 11 annually. These accidents are being caused because the pathways of elephants are being obstructed, and with the construction of electric fences, many animals are being pushed towards railway tracks or main roads.”
Tech solutions
Against such a backdrop, a high-tech attempt at preventing train collisions involving elephants is currently unfolding, which involves the use of Artificial Intelligence (AI) and infrared cameras.
According to National Innovation Agency (NIA) Additional Chief Innovation Officer Dr. Sachie Panawala, the NIA has been fundamentally and critically involved in this project to deploy a system to prevent elephants colliding with trains.
On 1 March 2024, during the steering council meeting of the agency, NIA Chief Innovation Officer Prof. Ajith de Alwis had brought to its attention research that had been carried out and successfully trialled in this regard, but which had been abandoned without any implementation.
Following discussion with council members, a meeting had been convened by the Presidential Secretariat with the NIA co-organising the event, which had seen the participation of all stakeholders in order to discuss several matters related to the railways.
Stakeholders present at this meeting had included the Railways Department, the secretaries of the Transport and Environment Ministries, and the Centre for Defence Research and Development (CDRD).
University of Peradeniya Faculty of Engineering Department of Electrical and Electronic Engineering Professor Lilantha Samaranayake had presented his research on an AI-powered system to prevent elephant-train collisions at this meeting.
This AI-powered technology is reportedly able to detect wild elephants on the tracks and alert train drivers about the presence of an elephant up to 500 metres away, providing ample time to take necessary precautions.
The device has already received approval from both the Railways and Wildlife Departments.
“Right now, even amid thick darkness, we can see elephants within about 500 m. This was a big challenge since the trains need a very long distance in order to stop, which means that we have to detect the elephant well in advance, since otherwise they cannot stop even if an elephant is detected,” Prof. Samaranayake said.
“These aspects have been incorporated and accuracy was also recently increased to about 83%. After deploying these devices, we will be collecting the data while the system is in place and operating as we undertake field trials. The data gathered during the field trial will be used to improve the accuracy and we plan to do the field trial for about three months to see if there are any improvements.
“Last Monday (24), a meeting was held, attended by the Deputy Minister and Secretary of the Ministry of Environment, as well as top officials including the General Manager of Railways and the DWC Director General. They agreed that this project should go forward. Only four such units will be needed to cover the relevant area if this is successful – one already exists and funding will be required to replicate.”
The system involves an infrared camera which can cover shorter distances of about 50 m, while the required range is 500 m. Therefore, a video capture is obtained from the camera and is converted real-time into images, which are sent through an AI engine.
This engine is trained using previously captured data on the track, allowing for the identification of elephants despite the lack of a camera that can see a distance of half a kilometre. The driver is notified through a buzzer inside the engine.
Trial and error
This tech-based solution has been in the works for several years, facing obstacles due to funding, technical considerations, and other concerns.
According to Prof. Samaranayake, they have been working on this project since 2020 through two competitive grants from the IEEE Robotics and Automation Society and the Nvidia hardware acceleration programme, which had offered funds for a proof of concept and hardware, respectively.
“We prepared to do field trials back in 2021 and we did some with the help of the Sri Lanka Railways administration at that time. We held one field trial to identify the problems in our system and see how we could improve. The initial idea was to implement it in certain hotspots and receive feedback.
“However, subsequently the General Manager changed and the new GM at the time, who also happens to be the current GM, said that it would be difficult to provide security to such installations because there was a threat of hardware theft and suggested that the system be fixed inside the cabin.
“After developing the software and when it was time for installation, another issue arose as we had to drill the train compartment to take some wires out, which caused an insurance problem. At that point, the project ran out of funds and was unable to proceed.”
Dr. Panawala noted that following the recent collision, there had been a discussion in Parliament on this matter on 20 February. Incumbent Environment Ministry Secretary K.R. Uduwawala, who had been a council member of the NIA at the time the system was initially discussed, had informed the gathering which had included Deputy Minister of Environment Anton Jayakody regarding the NIA-supported system.
Thereupon, a meeting had been convened with all stakeholders, which included the Railways General Manager and all necessary parties including Prof. Samaranayake and the research team, which had demonstrated the operation of the unit.
“The Railways GM, in particular, expressed his support to install these units inside the trains. Only two trains run on the Batticaloa line, so if you install these units in those two trains, that is all that is needed. Since Prof. Samaranayake expressed readiness to install the units, it was decided that he would get the support of the Railways Department and launch the trial next week,” Dr. Panawala said.
Prof. Samaranayake noted: “In June last year, there was a request from the Presidential Secretariat for me to come and present the work. They subsequently decided to fund field trials, which started in December. Last week (24), we got permission from the Railways Department to go for field trials. I am planning to send the setup with the research engineer next week.”
Confirming this, Jayasundara said that the Railways Department had agreed to install the system, noting that if the trial was successful they would be able to install this system in all trains.
“We will be fixing this unit permanently to one train on the Colombo-Batticaloa line as a trial. We will proceed based on the success of the trial,” he said.
He added that the DWC, Forest Department, etc. had also prepared an action plan to mitigate elephant-train collisions, such as providing technological solutions, clearing the sides of the railway tracks, launching an awareness programme for railway staff, and changing train travel times.
This is not the only tech-based intervention underway to address this particular aspect of HEC. Even as recently as January, the DWC initiated the operation of GPS technology to manage the movement of wild elephants from several areas in Anuradhapura to the Wilpattu National Park.
A futile endeavour?
However, this project is not without its detractors. Centre for Conservation and Research Chairman Dr. Prithiviraj Fernando, whose research includes Asian elephant conservation and management, expressed scepticism, suggesting that the attempt was misguided and “fundamentally wrong”. He opined that early detection of elephants crossing the track was a futile endeavour.
Addressing the AI camera system, he said: “It may work in rare instances, but overall it will be completely useless because most of the elephants that are getting hit are those that are crossing the track because the track goes through the area they are living in. When they are feeding or moving around, they repeatedly cross the track. It’s not that elephants sleep on the track – I am not saying it cannot happen, but that is not common.
“The issue is, if a train is coming along the track and elephants are crossing 500 m ahead, even if the camera detects it and informs the driver, it is pointless because that train is not going to hit that elephant. If the elephant crosses the train 100 m or 10 m in front of the train, it will not get hit.
“It is the elephant that crosses the track just as the train comes that gets hit. The problem is when the train is going at over 75 km per hour and if an elephant crosses a few metres in front of it, even if the train sees it, it will not be able to stop, because when brakes are applied at that speed, the train takes about a kilometre to stop. By then the elephant will have been hit, the train will have gone off the track, and another disaster will have happened.”
According to Dr. Fernando, a more likely solution that can be tried immediately is for trains to reduce their speed.
“When you look at the tracking data, elephants use certain areas intensively, so if we stop the train going through the area whenever there are elephants close to the track, we will not be able to operate trains in these areas.
“The only solution to this that is practical, that can be done immediately, and is effective is for the trains to travel slowly. After identifying the areas that are high risk, trains have to go slowly through those areas. There is no other way.
“If the trains travel slowly, then the possibility of hitting an elephant is very much decreased and there is more time for the driver to apply the brakes and slow the train further, as well as more time for the elephant to move away,” he said.
While innovations such as AI-powered systems show potential to address Sri Lanka’s multifaceted elephant problem, there appear to be significant practical realities eclipsing their effectiveness. Nevertheless, the input of various stakeholders highlight the need for an urgent solution before recovery becomes impossible.
Wildlife underpass: A white elephant?
In a move to address the escalating issue of Human-Elephant Conflict, last year, the Ministry of Transport and Highways, Sri Lanka Railways, and the Surakimu Railway National Organisation initiated the construction of Sri Lanka’s first wildlife underpass in Kasikote, Galgamuwa.
The construction of the wildlife tunnel under the railway line commenced on 1 March 2024 in Gatadivula-Kasikote of Galgamuwa under the directions of then Minister of Transport and Highways Dr. Bandula Gunawardena.
The official statement at the time notes that a policy decision was taken to construct this wildlife tunnel under the Maho-Omanthai railway rehabilitation project, which was being done with the financial assistance of an Indian line of credit.
However, questions remain on the feasibility of the project.
According to National Innovation Agency Additional Chief Innovation Officer Dr. Sachie Panawala, experts in the field had not been consulted as to the feasibility of this project.
“Elephant passes are not a new thing; they have been constructed in other countries like Kenya and even Kerala. According to experts like Prof. Ashoka Dangolla, however, such underpasses have built a railway or a highway over the elephant path, not the other way around – not under the already existing highway or railway.”
Similarly, Centre for Conservation and Research Chairman Dr. Prithiviraj Fernando, although expressing approval of the concept, pointed out the practical complications, especially the high expense.
“The underpass for wildlife is excellent. If we can construct those, the problem will be solved. However, the issue is that we have to construct many such underpasses. Although it is an excellent initiative, as far as I know it has not been finished.
“We will also likely need an underpass in areas that elephants use intensively, perhaps every 500 metres or so. Therefore, there are issues in terms of how practical it is, whether we can afford it, and whether we can do it,” he observed.
He further noted the practical obstacles to implementing underpasses as a solution.
“If these underpasses are located in the correct places, then we can expect that elephants will use them. If there are sufficient underpasses, we can also install some fences in the intervening areas and try to guide them. However, this has to be done very carefully, because if there is a stretch of 10 km where elephants cross the track and just one underpass, many of the elephants will try to cross from the other fenced areas.
“If the underpass is built in the middle, elephants will sometimes have to walk 5 km to get to the underpass, which they probably will not do. But if there are underpasses every 500 m or so, then yes, it might be possible.”
According to him, such an initiative will take time, a great deal of effort, and even more importantly, significant funding.
According to General Manager of Railways J.I.D. Jayasundara, the underpass in Galgamuwa is currently under construction. However, he was unable to provide a definite timeline for completion, noting that it was being carried out under a project through an Indian credit line, with delays currently being caused given issues with obtaining the funding due to International Monetary Fund (IMF) conditions.
He however asserted that the construction would be completed this year.
Bar graph
Elephants killed by trains
2018 – 16
2019 – 8
2020 – 3
2021 – 6
2022 – 14
2023 – 24
2024 – 11
Jan.-Feb. 2025 – 9
(Source: DWC)