For commonly used Unmanned Aerial Vehicles (UAVs) multirotors, motor thrust estimation techniques do not accurately describe the actual thrust produced by the propellers, as they rely on the motor command or the motor angular velocity and a rough aerodynamic model that relates the latter with the thrust produced by the propeller. In various cases, such as, e.g., Electronic Speed Controllers (ESCs) communication loss, flight in presence of wind, ground effect, and propeller failure, the actual thrust can differ significantly from the one estimated using the model. Such a problem is particularly limiting in the case of aerial robots that are employed in physically interactive tasks such as inspection and maintenance.
To tackle this limitation, in RAM we developed a 3D printed thrust sensor that is embedded inside the arm of the aerial robot. The sensor is printed using electrically conducting PLA Proto-Pasta material, together with the arm of the drone. This offers the flexibility to upgrade the sensor and arm designs easily to meet specific requirements and eases the process of integration of the sensor itself. The current sensor has been tested and validated on a single drone's arm on a static test-bench.
In the frame of the Aerial-CORE project, this master thesis assignment will consist of extending the work done with the current sensor to a fully flying multirotor platform. The practical objective of the thesis is to perform successful flights with a new multirotor that will embed the sensor in every arm. The scientific objective is to find a way to exploit such new available sensors to improve the performances when in physical interaction with the environment, e.g., during inspections, as it is done for torque-controlled robotic manipulators.
Practically, the task will include:
- Design and prototyping of the frame, integrating the sensor (sensitivity, arm deflection, vibrations, etc), with the goal of achieving a reliable force measurement while minimizing undesired or challenging aspects such as vibrations and deflection
- Modelling and Identification of the aerial platform with flexible arms, with the goal of including the flexibility in the model, finding the least amount of parameters needed to represent the system in a faithful way and identifying them.
- Control design and Flights experiments, with the goal of performing physical interaction tasks using the additional force measurements provided by the sensor in the control law to improve the task performances.