Hi Everyone! I (and my team) are new to the WildLabs network so we'd like to post an early-stage project we've been working on to get some feedback! 

Summary

The utilization of acoustic data collection stands at the forefront of modern conservation methodologies, particularly with the integration of wearable technology. This sophisticated approach yields invaluable insights into the nuanced dynamics of wildlife behaviors, inter-species interactions, and the indispensable ecological functions within their respective habitats (Oswald et al. 2022, Robbins and McCreery 2003). This has led our team at the MIT Media Lab to develop CollarID, a device that tracks animals and their environment synchronously, as a new tool for wildlife biologists to answer fundamental questions about the interactions of large predators over its one-year deployment. By combining environmental data (i.e., acoustics, temperature, humidity, and gas) from CollarID with animal movement and behavior data (via GPS and accelerometer) from both CollarID and widely-used GPS collars, researchers can gain valuable insights into endangered populations. 

Attaching bioacoustic monitoring devices to an animal, especially via existing commercial GPS collars, allows us to capture a wealth of supplementary data on an individual and their surroundings. With this device, our aim is to gain insight into two canid-applications: unsupervised monitoring of dog behavior and the stimuli they encounter for both domesticated dogs and African Wild Dogs. Our end goals surrounding African Wild Dogs include collecting information on habitat use, proximity to and interactions with humans, population connectivity, and overall interactions, to address primary conservation concerns of wild dogs (Mills et al. 1998, Woodroffe and Ginsberg 1999, Creel and Creel 1998). To do so, we’re leveraging sensing modalities that collect information on ambient temperature, humidity, air quality, acoustics, and movement. While our current focus is on dogs, this device can be attached to other similar-sized or larger animals for related data collection. 

Our CollarID prototype features a water and dust resistant microphone, temperature, humidity, gas, and inertial sensing systems enclosed in a ruggedized housing. There are two inertial systems, one 6-axis system (magnetometer, accelerometer) for high frequency sampling, and one 3-axis system (accelerometer) for long term and low-power activity detection used to trigger other subsystems. The device features an SD card and a low-power microphone sampling design that virtually shuts off the entire system while sampling, only waking up when saving data to the SD card. The device is designed to be configurable to support both long- and short-term applications. The system also features a removable solar circuit to be used where solar harvesting is possible, and includes a LoRa radio for long distance communication and status updates. We are currently developing and testing the system, with plans to complete system trials later this year. Our current work is focused on evaluating the sensing subsystems and then optimizing for weight and size with the first prototype aiming to be less than 100 grams and smaller than 35 x 35 x 75 mm^3. 

We are initially testing our system on domesticated dogs, including both service dogs and household pets. Environmental data, from a combination of acoustic, activity, and environmental sensors, can give researchers insights into patterns of stress in domesticated dogs, and inform technological improvement before deploying CollarID on African Wild Dogs in the wild. In this pursuit, we plan to pair our wearable sensors with existing chew toy sensing to further classify dog behaviors and stress signals (Krichbaum et al. 2023, Ramey 2024).

We have gained a lot of insight over the years by engaging with the broader community in constructing technologies for wildlife monitoring (e.g., BuzzCam). Please ask us any questions on our system development, or let us know how you could apply this technology to your research. We would love your feedback!

Team Members: Patrick Chwalek, Kasim Rafiq, Sophie Montague