Oral Presentation- Symposium 12th International Mammalogical Congress

A scalable approach to wild animal monitoring using distributed autonomous sensing systems (#162)

Ashley Tews 1 , Philip Valencia 1
  1. CSIRO, Pullenvale, Queensland, Australia

Camera traps have enabled significant insights into wild animals' lives through the collection of image-based information which now routinely include high quality day and night imagery due to continuing advancements in camera technologies. However, the effective range of a camera trap is restricted by its short range motion sensor and highly manual and logistical burden to recover and analyse data. This typically limits their utilisation to small areas of coverage, or the need for attractants within the sensor's range. Through our research into humanely deterring vertebrate pest animals, we need to address issues of scalability (coverage area), detection range, and in situ animal classification for real-time pest management. Our approach employs a combination of inexpensive, low-power, ‘detection nodes’ with motion sensors and wireless radio communications capability, operating in collaboration with a ‘‘supervisor” that has high quality thermal and colour cameras and also has wireless communications capability. This combination provides low power, scalable coverage for triggering long range, high quality thermal and colour video recording. Additionally, this allows for onboard, in situ animal classification. Examples where our approach can be utilised include game trails, farmland, and landscape-transition boundaries, covering larger areas than practically possible with the stand-alone motion based sensor approach. Further to the discussion of the potential benefits and challenges of our approach, we present results from early instantiations of our system deployed in peri-urban, bush and agricultural domains.