In this project, we are developing a novel bio-inspired acoustic sensing-processing system to detect koalas. The proposed system solves the critical issues faced by conventional microphones when mapping Koala populations, enabling more accurate and efficient estimation of Koala numbers - while avoiding the generation of large, impractical, amounts of data. Most importantly, this will complement the data captured by existing drone surveys done by The Department of Planning, Industry and Environment (DPIE).
Remote ecological monitoring often faces challenges due to power constraints, data overloads, and the requirement for long-term deployment without regular maintenance. This project proposes an integrated system employing machine learning and neuromorphic engineering, designed to efficiently process acoustic data and enable long-lasting, autonomous operation in remote locations.
Neuromorphic systems, inspired by the human brain, excel at real-time, low-power processing. Leveraging neuromorphic hardware such as event-based sensors and neuromorphic processors, we aim to develop a system that consumes power only when processing changes in the acoustic environment, thus enabling long-term deployment.
Machine learning will play a crucial role in the analysis and interpretation of the captured data. The system will be trained to identify and tag specific acoustic events that indicate particular ecological phenomena, thereby reducing the volume of data that needs to be stored or transmitted and preventing user overwhelm. This strategy allows for focused monitoring of ecological changes without the need for human intervention.