AI-powered monitoring of livestock health and welfare using multimode sensor data
Supervisors: Reza Arablouei (CSIRO), Greg Bishop-Hurley (CSIRO) and Rachael Rodney Harris (ANU).
Maintaining animal health and welfare is crucial for productivity and sustainability of livestock agriculture. Manual observation of animals to monitor their health and welfare is prohibitively laborious, especially when their number, or the area they spread over, is large. Multimode sensor data, e.g., inertial measurements and position, collected via wearable devices such as collars or ear tags can be used to infer animal behaviour, movement patterns, and interactions with the environment or other animals.
In this project, the PhD candidate will develop new data analytics and machine learning algorithms that enable monitoring livestock health and welfare. Utilising lower-level information extracted from multimode sensor data such as behaviour, activity, or social interactions, the candidate will analyse existing sensor data to gain insights into hidden short- or long-term patterns that can guide development of new models and algorithms, which relate key health and welfare parameters to multimode sensor data.
The developed technologies will help automate livestock health and welfare monitoring, and ultimately manage animal health and welfare more effectively. This will improve the efficiency of livestock agriculture, reduce its environmental footprint, and make it more sustainable, in addition to promoting animal welfare and regulatory compliance.
Developed algorithms will be considered for implementation on embedded systems of wearable devices to collect further data to refine the learned models. The theoretical properties of the proposed solutions will also be analysed.