Tamil Nadu: AI intervention to address human-elephant conflict in Nilgiris

Chennai: Gudalur in the picturesque Nilgiris district has been one of the hotspots for man-elephant conflict in Tamil Nadu leading to human casualties, crop damage, and property destruction. After several conventional mitigation methods failed to prevent the conflict, the Tamil Nadu Forest Department has now turned to Artificial Intelligence to address the issue.

The department’s first initiative using AI-early warning system in Madukkarai in Coimbatore district turned successful after no deaths of elephants were reported on the 13-km stretch on the Pothanur (Tamil Nadu)-Palakkad (Kerala) railway section since November 2023 when the new system was implemented.

At least 11 tuskers died after being hit by speeding trains in the above section between 2008-2021. The system has enabled nearly 6,000 safe elephant crossings in the past two years.

The dedicated command and control centre at Nadugani Range in the Gudalur Forest Division, which was inaugurated last week, will use 46 AI enabled thermal cameras to detect wildlife movement early and warn people with real time alerts.

Man-elephant conflict has been a regular affair in Gudalur area since it falls within the migratory route of the jumbos from Tamil Nadu to Kerala and accidental encounters take place when they trespass into habitation areas.

“Of the 46 cameras, 34 are AI-enabled IR cameras, while the remaining are AI thermal cameras. These 46 points where the cameras are placed have been identified using 10 years of data of where elephants were detected in the past. Once the camera identifies an elephant, it triggers an alert to the Command Centre,” N Vengatesh Prabhu, DFO (Gudalur), told DH.

The alerts are transmitted instantly to field staff through the command and control centre through 24 speaker systems positioned near human settlements as well as through SMS notifications.

Prabhu added that even cameras are capable of transmitting signals, triggering direct alerts to people of the area, whose mobile numbers are feeded in the system, and the Forest Department staff.

“If the accuracy (of elephant spotting by the AI cameras) is 92 per cent, it directly sends alerts. If the accuracy is less than 92 per cent, the alert is sent to the command and control centre where the staff will verify the image and trigger the alert manually,” he said.

The DFO said an alert will also be issued to nearby patrolling teams which will jump into action immediately and send back the elephants to the forest.

To reduce the response time from the current 20 minutes to half in about an year, the department will follow the predictive modelling to find out the most vulnerable areas and station the patrolling teams directly there.

“We hope to achieve this in about a year’s time. This is an integrated approach to conflict management in arguably the biggest hotspot for human-wildlife conflict in the state,” he added.

The division has also been provided with two advanced drones with thermal imaging capabilities for monitoring elephant movement and to support field-staff during elephant-driving operations.