skip to content

Department of Computer Science and Technology

We've all read the urban myths about how evil Artificially Intelligent robots are going to take over the world. Now meet 'Terra' – an AI that researchers hope will help us save the planet by mapping the terrestrial life on it. 

Terra is the brainchild of a group of Cambridge University computer, climate and conservation researchers who have been working together for several years in the Cambridge Conservation Initiative.

When fully developed, this AI tool will give governments, businesses and scientists radically improved insights into what the climate crisis is doing to our planet.

Armed with its modelling and predictive powers, they can then make much more effective decisions about how we can conserve species, restore the climate and protect biodiversity globally while still supplying food, energy and water to the world’s population.

The scientists – gathered from this Department, from the Departments of Plant Sciences and Zoology and from the UN World Conservation Monitoring Centre (UNEP-WCMC) – have just been given the green light to start work developing 'Terra'.

UK Research and Innovation, the largest public funder of research and innovation in the UK, announced today (2 September 2024) that it will fund this project as part of a programme to encourage exciting new interdisciplinary research. 'Creating foundation systems for environmental planetary intelligence' is one of 36 projects receiving funding today under a new, cross-Research Council funding scheme to unlock new research, approaches and methods. 

This research is focused entirely on understanding, mitigating and reversing the extinction crisis.

Prof Anil Madhavapeddy

 

But in fact, the researchers behind the Terra project have already been collaborating for some time in a range of areas that harness the power of computer science to tackle major planetary challenges. And this is vital, according to its Principal Investigator.

"The loss of biodiversity globally is accelerating, as is the deterioration of critical ecosystems and the release of greenhouse gases," says Anil Madhavapeddy, Professor of Planetary Computing here, who will be leading the project. "We've got to radically reconsider how we respond and to do that, we need better ways to measure the impact of human activities on the planet."

To succeed, he adds, we must develop new and innovative AI systems (such as Terra) to model and predict what’s going on in our world because "we need to move five times faster in conserving the planet if we’re going to meet our goals under the Kunming-Montreal agreement".

AI, of course, is already being widely employed to help scientists model and understand climate change and biodiversity loss. But the solutions are not keeping pace with the urgent need for them. "Current models aggregate fragmented and misaligned data derived from expert opinions and field inventories," Anil says. "And most digitisation efforts cannot model the hugely complex dynamics involved. Nor can they address these problems at scale.

"This is why, at the moment, policymakers and the private sector cannot accurately identify the potential impact of critical decisions." This is where the interdisciplinary team of Cambridge scientists comes in. Alongside Anil Madhavapeddy they include Srinivasan Keshav, Robert Samson Professor of Computer Science here, Andrew Balmford, Professor of Conservation Science in the Department of Zoology, and Professor David Coomes, Director of the Cambridge Conservation Research Institute.

Professor Neil Burgess, Chief Scientist at the UN World Conservation Monitoring Centre, is also co-leader on the project, while our two recently appointed Planetary Computing Fellows, Dr Sadiq Jaffer and Dr Michael Dales, will also be driving technology development. Together, they head a team that will create Terra – a predictive AI model of the world – by using their expertise in combining extensive earth observation data (from satellites and drones) with terrestrial data (from sensor networks, citizen science and habitat maps) and using self-supervised AI training.

"It's our aim that Terra will be able to make predictions for a great many plant and animal species on which we currently have only sparse data," Neil Burgess says. "We hope it will also enable monitoring of biodiversity at much higher resolution and accuracy than is currently possible."

Another key aim is to link data on short- and long-term changes in species' habitats so Terra can model how human actions impact extinction risks, and where and how best these might be mitigated. "We'll use Terra to build a suite of critically important predictions about life across our planet," says Anil. Using Terra, the aim is to give governments and businesses such accurate and detailed assessments of how human activities – like farming, processing and distributing food – are impacting on nature that it enables them to make the best conservation decisions.

"This research is focused entirely on understanding, mitigating and reversing the extinction crisis," Anil adds.


Published by Rachel Gardner on Monday 2nd September 2024