Oxford researchers revolutionise 3D wildlife tracking with WildPose

A groundbreaking new technology developed by researchers from the University of Oxford and their international collaborators is set to transform how scientists study animal movement in the wild.  

Andrew Markham, Professor of Computer Science, together with DPhil students Sangyun Shin and Qianyi Deng, collaborated with Amir Patel (UCL, formerly University of Cape Town), who pioneered the project, and PhD student Naoya Muramatsu (University of Cape Town) to develop WildPose, a pioneering system that combines LiDAR with high-speed telephoto imaging. This system allows researchers to capture the biomechanics of wildlife remotely from over 100 metres away, offering new insights into animal behaviour in their natural habitats.  

With support from Google Research, the team successfully deployed the system in Kgalagadi Transfrontier Park, South Africa. There, they recorded a range of species in motion, from the footfalls of red hartebeest to the breathing rhythms of resting lions. The results offer remarkable insights into animal biomechanics, without the need for invasive tracking methods such as GPS collars. 

For decades, researchers studying animal movement have relied on laboratory settings or expensive GPS tracking to analyse behaviour. However, these approaches have limitations: laboratory settings do not fully capture natural movement, and GPS data alone lacks the fine detail needed to understand biomechanics at a high temporal resolution. 

LiDAR, a laser-based 3D scanning technology, offers a promising alternative. By mapping the terrain with extreme accuracy, LiDAR creates a detailed three-dimensional representation of an animal’s environment. The challenge was integrating this technology with high-speed imaging to capture animals in motion from a distance. 

To build WildPose, the team needed a LiDAR system capable of producing high-resolution 3D scans to complement a high-speed (170 frames per second) high-definition telephoto camera. Muramatsu and Patel spent months refining the hardware and software, ensuring the system could track animals moving at varying speeds and distances while maintaining accuracy. 

After extensive testing, which included filming Muramatsu himself walking in circles to verify precision, the team took WildPose to South Africa for its first field trials. Mounted on a 4x4 vehicle, the system proved highly effective at capturing the movements of animals in their natural environment, despite challenging conditions such as extreme heat, vibrations, and the need for vast amounts of data storage. 

During field trials, WildPose successfully recorded a range of species, including giraffes, springboks, jackals, and martial eagles. In Oxford, Andrew Markham, Qianyi Deng and Sangyun Shin helped analyse the recordings and were able to measure the height of a 5.5-metre giraffe at a distance of 90 metres and reconstruct the gait of a red hartebeest, determining that it was walking at a precise speed of 0.844 metres per second. The team even recorded the subtle chest movements of a reclining lion, mapping its breathing at a rate of 80 breaths per minute - offering insights into the animal’s physiological state without any physical interference. 

‘WildPose is the culmination of a number of years of discussions that Amir and I had about how we could revolutionise the way wildlife can be tracked and monitored in 3D with minimal disturbance. This international collaboration has been a brilliant experience, especially the opportunity to work with talented researchers from my alma mater. I believe WildPose is a first step towards an exciting new era of rich 3D data from the wild.’ – Professor Andrew Markham. 

Beyond its applications in biomechanics, WildPose has the potential to revolutionise wildlife conservation and health monitoring. The ability to observe animals remotely at such a high level of detail could provide valuable data for understanding the impact of climate change, monitoring endangered species, and even informing engineering applications inspired by natural movement. With future deployments planned, the researchers hope to refine WildPose further and expand its use in ecological studies.