Recently UAVs have gained lots of attention in both research facilities and the public. UAVs proved to have promising capabilities for applications such as inspection and logistics.
For instance, in 2019 a company called Zipline focused their attention to deliver medicine in far reaching countries with limited infrastructure by using a fixed wing UAV. Also NS ,the Dutch railway operator, is going to use UAVs to inspect for maintenance on their buildings. Furthermore, A Concordia University research team is arguing that drones are the way of the future for monitoring and combating forest fires in Canada's vast boreal forest . In summary, as the technology of UAVs matures, more applications for societal benefit will be possible.
In 2016 the Robotics and Mechatronics group started working together with 20 Dutch and German small-and-medium sized enterprises on the SPECTORS project (Sensor Products for Enterprises Creating Technological Opportunities in Remote Sensing). One of the productgroups of the SPECTORS project is the employment of drones for physical interaction with the environment. Successful adoption of these features would enable UAVs to engage in application where interaction is necessary.
Up to now, work has been done to develop the fully-actuated hexarotor. This UAV has tilted propellers such that the UAV can be actuated in every DOF. This enables the UAV the more easily apply a force in a single axis without the need of additional pitching angles. Next to this the UAV also uses a port-Hamiltonian framework with and additional energy tank. In combination with an impedance controller the energy tank is being used such that the UAV will remain passive while flying. This will prohibits the system from becoming unstable while interacting with an object.
Along with this passivity-based control, the UAV makes use of a localization system and a disturbance observer. Currently the localization is being achieved by the use of "MoCap" a motion capture system. This system makes use of different cameras in a drone testing lab to monitor the pose of the UAV. It acts as IMU bias correction as well as ground truth while the UAV is flying in the lab. The disturbance observer acts alike to a low-pass filter and it enables the UAV to identify disturbances while performing slowly changing manoeuvres.
By using the previous described methods, interaction between the UAV and a surface has successfully been achieved. However, during the project phase in the time period 2019-2020, the UAV should start interacting with objects in an outdoors environment. This means that first of all, the MoCap system can not be used anymore and secondly the wrench observer will be too primitive.
This last statement is true because the current wrench observer does not distinguish between interaction wrenches, (wind) disturbances and modelling errors. Furthermore due to its low-pass nature the observer is limiting the UAV to low frequent behaviour.
This assignment will focus on developing an estimator, for the fullyactuated UAV, that is able to distinguish between the different wrenches such that a robust interaction can be achieved.