AI for the environment
While artificial intelligence (AI) has captured public attention globally the use of AI tools is not entirely new; AI and machine learning (ML) applications have been used in environmental sciences for decades. The true extent of the impact that AI will have on human society is as yet unclear. As awareness of its potential grows in the public conscience, calls for conversations around its risks, challenges and opportunities have come forth from global and local bodies. Our office is considering how AI might be useful for improving outcomes for our natural environment.
What applications are already in place?
Examples of useful AI research and technology that supports positive environmental outcomes are already available; we describe some examples here.
Remote sensing
Recognising that the standard tools used for ecological assessments were limited in their ability to provide regular, comprehensive information about how our national biodiversity was faring the team at Eco-index saw that a new approach would be useful. Through the application of AI tools to high resolution satellite imagery Eco-index are aiming to understand how native ecosystems change with time. These tools will help to inform our understanding of wetland ecosystems resulting in better management of wetlands and providing a mechanism to quantify restoration works.
Wallaby trapping
The 2023 “Solve for Tomorrow” competition was won by Rotorua local, Cameron Moore, who created a humane trap for wallabies. While mountain biking in the Whakarewarewa forest, Cameron spotted a wallaby eating small shoots from the forest floor. With the intention of enhancing the forest biodiversity, Cameron set about designing an AI enhanced wallaby trap. A camera connected to the trapping box was programmed so that a motorised door would open if the AI detected that a wallaby was nearby. Because the trap utilises AI, it is able to distinguish between wallabies and other traffic that is common to the Whakarewarewa forest including dogs and small children.
TAIAO
TAIAO is a collaborative effort between the Universities of Waikato, Auckland, Wellington (Victoria) and Canterbury, Beca and MetService. Recognising that good data are essential to research, understand, and set policy for effective management of our natural environment, the team at TAIAO intend to develop new machine learning methods that are tailored to process environmental data gathered from our New Zealand context. TAIAO plan to codesign their work with iwi, industry, and government with the vision of harnessing data science to preserve our natural resources. In parallel with developing open-source software, TAIAO also plans to build New Zealand’s environmental data science capabilities through the delivery of workshops, undergraduate, and postgraduate research projects. The Aotearoa Species Classifier app (pictured) is an example of a publicly-available tool which uses machine learning to help people identify around 11,000 species of New Zealand flora and fauna.
Video resources
The team at Zero Invasive Predators (ZIP) have used AI software, coupled with a thermal camera, to identify predators and provide notification of detection.
Researchers from Queensland University of Technology (QUT) are using AI to analyse infrared imaging in efforts to identify koala populations.
Last edited on: 2nd May 2024