The prototype of the Elephant-Train Collision Prevention System, a product of a collaborative project initiated and conducted by a group of DEEE members, gained media attention after the recent unfortunate elephant-train accident at the Galoya-Habarana rail track. 

This initiative to protect both wildlife and rail transport introduced an advanced AI-powered sensor system aimed at preventing train-elephant collisions. This visionary project, initially funded by the IEEE RAS-SIGHT (Robotics and Automation Society Special Interest Group on Humanitarian Technology) through its 2019 Call for Humanitarian Projects, is set to revolutionize railway safety in regions where elephant crossings pose a significant risk.

In the initial phase of the project, the concept was to strategically deploy sensor units designed to monitor key “hot spots” where elephants frequently gather and cross train tracks. Using cutting-edge AI and sensor technologies, the system continuously scans the environment and transmits real-time alerts to an onboard computing unit. These alerts enable train drivers to take immediate action, significantly reducing the risk of accidents and fostering a harmonious coexistence between infrastructure and wildlife.

Building on this success, the second phase of the project introduced an advanced in-cabin detection system, designed specifically for train drivers. Developed with funding from the prestigious NVIDIA Applied Research Hardware Accelerator Program, this AI-driven unit is capable of detecting elephants both near and far—up to 500 meters ahead of a moving train. Leveraging state-of-the-art machine learning techniques, the system ensures reliable elephant detection under diverse lighting and environmental conditions, operating seamlessly both day and night.

This initiative is not only a triumph in engineering but also a testament to the power of collaboration and academic innovation. Several final-year undergraduate research projects have also contributed valuable insights, exploring additional sensor modalities and refining detection techniques to further enhance the system’s effectiveness.

By merging AI with real-world conservation efforts, this pioneering project exemplifies how technology can be harnessed for the greater good—protecting wildlife, ensuring safer train operations, and setting a new standard for responsible and sustainable transportation.

DEEE would like to congratulate the research team (Prof. Lilantha Samaranayake [PI], Prof. Kithsiri M. Liyanage, Dr. Nalin Harischandra, Dr. Tharindu Weerakoon, Eng. Shanaka Ramesh Gunasekara, Eng. Tuan Azzam from UOP; Prof. Gamini Dissanayake [PI], Dr. Ravindra Ranasinghe, Dr. Maleen Jayasuriya Collaborators from University of Technology, Sydney) and wish them success in their future innovations and investigations.

Publications:

• -A convolutional neural network-based early warning system to prevent elephant-train collisions, In 2021 IEEE 16th International Conference on Industrial and Information Systems (ICIIS2021)
• -Enhanced Frequency Domain Analysis for Detecting Wild Elephants in Asia Using Acoustics, In 2023 IEEE 17th International Conference on Industrial and Information Systems (ICIIS2023)
• -A vision-based early warning system to prevent elephant-train collisions, In International Conference on Wildlife Technologies and Wildlife Conservation (ICWTWC2021), London, UK
• -Elephant train collision prevention system, In workshop on Humanitarian Robotics, IEEE International Conference on Robots and Systems (IROS2020), Las Vegas, USA, 09 November 2020 (ONLINE) 

Media:

• - New device to prevent elephant-train collisions (Newswire)
• - අලි අනතුරු අවම කරගන්න AI තාක්ෂණයෙන් විශේෂ උපකරණයක් (ITN News)
• - Camera developed to save elephants from being hit by trains (DailyMirror)
• - වන අලින් දුම්රියේ ගැටීම අවම කිරීමට කෘත්‍රිම බුද්ධියෙන් උපකරණයක් (UOP- Public Relations Division)
• - AI Based solutions for Elephant-Train Collisions in Sri Lanka (EEES)

Final Year Projects:

• - Airborne Sound for Elephant Localization in the Wild
• - Seismic Waves for Elephant Localization in the Wild
• - Action Prediction of Wild Elephants using vision-based Deep Learning