Machine Learning for Remote Sensor Placement with Multi-modal Aerial Robots

Drones today are mainly used to do aerial photography without the capability to do interaction with the environment. Moreover, they usually only use vision based technologies that does not allow them to collect sufficient information about the environment to inform agricultural decisions or validate ecological models.

This project will investigate how drones can be used to deposit sensor networks in the environment using aerial deployment methods. The technological development will include machine learning control systems to select applicable sensor locations and to ensure precise positioning of the sensors in the environment. The work will be validated in outdoor terrain as well as in complex constrained environments (such as in forests).

The PhD student will focus on novel control and hardware solutions as well as data analysis approaches to identify environmental parameters with the developed aerial systems. This capability will open a new field within both, engineering and sensor networks as well as in ecology and environmental monitoring where the collected data is of very high value. The collected data can also be validated against traditional methods of manual sensor placement on trees that are performed with abseiling etc which are much more dangerous and costly compared to the proposed solution.