Taiwan group proposes AI-driven system to predict and warn of roadkill hotspots

The citizen-science initiative Taiwan Roadkill Observation Network has drafted a proposal that pairs AI modelling with a live warning system to curb wildlife-vehicle collisions across Taiwan. By integrating its own database of animal deaths with data from the Taiwan Biodiversity Network, public hotline reports, Ministry of Transportation and Communications accident figures, road network maps, traffic volumes, micro-climate data from the Central Weather Administration and Google Street View imagery, the algorithm will pinpoint road stretches with a high likelihood of strikes.

The system will activate changeable message signs equipped with infrared and light sensors to alert drivers when animals approach, beginning with a solar-powered prototype to be trialed on roads managed by the New Taipei City government. The initial pilot targets the endangered pangolin, with plans to expand to leopard cats, land crabs and breeding frogs. The proposal was entered into the 2026 Presidential Hackathon and advanced to the final round, where winners could receive state support for nationwide deployment. TaiRON, launched by the Ministry of Agriculture’s Taiwan Biodiversity Research Institute in 2011, already hosts over 330,000 records contributed by more than 36,000 volunteers.