Unmanned Aerial Vehicles (UAVs) have a wide range of civilian and military applications, such as, search and rescue, border management, mapping and surveillance. These applications often require UAVs to operate autonomously for extended periods of time, and over extended distances. As such, these UAVs require the use of control systems which are able to support complex requirements. These control systems generally consist of an onboard system to handle a wide variety of sensors in the UAV, and to perform rapid data integration for meaningful processing. A control mechanism is also needed to control the actuators.

In this project, the NeuraBASE network model was used as reinforcement model of an autonomous controller for a UAV. This network model represents a learning hierarchy of interconnected neurons capable of storing sequences of sensor and motor neuron events. The model was evaluated using experimental scenarios simulated on the STAGE robotics simulation software, involving navigational control towards a target.


A UAV pursues a target (blue cap).


3 UAVs pursuing 30 targets, each UAV can be controlled by a common or an independent NeuraBASE.


A UAV navigates towards a target while avoiding obstacles. 


A UAV landing.