Background
Bidirectional propellers allow a multirotor to generate thrust in both directions along each propeller axis. They are therefore important for fully actuated aerial robots and accurate physical interaction.
Recent theoretical work has revealed a fundamental limitation of bidirectional propellers. [1] Their thrust depends signed-quadratically on rotor speed. Reversing thrust therefore requires the propeller to pass through zero speed, where the ability of bounded motor torque to change thrust vanishes.
A thrust reference that crosses zero with nonzero slope cannot be tracked exactly using bounded motor inputs. This limitation is intrinsic to constant-pitch propellers. It cannot be removed by improving the low-level motor controller alone.
A related limitation appears in multirotor control allocation. In a nonredundant fully actuated multirotor, reversing one propeller temporarily removes one task direction. In a redundantly actuated multirotor, suitable control allocation can avoid or repair some problematic reversals by exploiting alternative actuator states that generate the same wrench. [2]
These aerodynamic limitations can be obscured in conventional sensorless brushless-motor systems. Such systems can already perform poorly around zero speed because rotor motion is estimated from the back electromotive force induced in the unpowered motor phase. Separating this implementation issue from the intrinsic thrust-reversal limitation requires direct rotor-position sensing.
Objective
The objective of this thesis is to demonstrate the fundamental thrust-reversal limitation on a real multirotor and develop a control-allocation method that overcomes it through redundant actuation.
The work will progress from the development of a sensor-based bidirectional propulsion unit to actuator-level experiments, multirotor integration, control design, and flight validation.
Main activities
- Develop a bidirectional motor-drive-propeller unit with direct rotor-position sensing.
- Select and integrate a magnetic encoder or equivalent rotor-position sensor that remains effective at low and zero speed.
- Implement reliable sensored motor commutation and closed-loop speed control in both rotation directions.
- Demonstrate that the propulsion unit can cross zero speed and reverse direction without relying on sensorless back-electromotive-force estimation.
- Design an actuator-level experiment for tracking thrust references with different behaviours around zero thrust.
- Measure rotor speed, thrust, motor current, voltage, and tracking error during thrust reversal.
- Test whether the measured current, voltage, and thrust responses agree with the predicted reversal-order limitations.
- Produce multiple sensor-based bidirectional propulsion units for integration into a multirotor.
- Test how individual propeller reversals affect wrench generation and tracking on a complete multirotor.
- Design a control allocator that exploits redundant actuation to avoid dynamically unrealizable or poorly conditioned thrust reversals.
- Implement the allocator on the multirotor and validate it in flight or physical-interaction experiments.
- Compare the proposed allocator with a conventional pseudoinverse-based allocation method.
Expected outcomes
- A sensor-based bidirectional propulsion unit capable of reliable operation through zero rotor speed.
- An experimental setup that separates motor-commutation limitations from the intrinsic aerodynamic limitation of thrust reversal.
- Experimental validation of the predicted relation between thrust-reference regularity and the required motor input.
- A complete multirotor platform equipped with sensor-based bidirectional propulsion units.
- A differentially realizable control-allocation method that uses redundant actuation to avoid problematic propeller reversals.
- Experimental evidence showing whether the proposed allocation method improves wrench tracking and physical-interaction performance.
Project structure
- The work is organized into successive stages.
- Completion of the sensor-based propulsion unit and actuator-level experiments already constitutes a substantial contribution.
- Multirotor integration and control-allocation experiments form the subsequent stages if progress permits.
Required or preferred skills
- A solid background in control, mechatronics, and electrical drives.
- Experience with embedded programming in C or C++.
- Interest in brushless-motor control, magnetic sensing, multirotor modelling, and experimental robotics.
- Experience with field-oriented control, electronic-speed-controller firmware, MATLAB/Simulink, Python, or ROS is beneficial.
Supervisors
- Antonio Franchi
- Ahmed Ali
- Chiara Gabellieri
References
- [1] Exact Thrust-Reversal Limits of Bidirectional Propellers under Bounded Motor Inputs, A. Ali, C. Gabellieri, and A. Franchi, arXiv:2608.06991. Preprint.
- [2] Differential Realizability of Static Control Allocation in Multirotors: An Impossibility under Nonredundant Full Actuation and a Pseudoinverse Obstruction under Redundant Actuation, A. Franchi and M. Mizzoni, arXiv:2609.13588. Preprint. [arxiv.org]