Gimballed Rotor Mechanism for Omnidirectional Quadrotors

AI summary

Overview

Research area: Aerial robotics, specifically mechanical design and control of omnidirectional multirotor uncrewed aerial vehicles (MRUAVs).

Technical level: Intermediate. The mechanism design is described in accessible terms, but the control allocation section assumes familiarity with linear algebra, control effectiveness matrices, and the PX4 Autopilot firmware architecture.

Scope: The paper designs, models, builds, and flight-tests a gimballed rotor mechanism that adds independent tilting to each rotor of a conventional quadrotor, along with the PX4 firmware modifications needed to control it.

What This Paper Is About

Conventional quadrotors are underactuated: because thrust always points along the body's vertical axis, position and attitude control are coupled, and the vehicle must tilt to move. Omnidirectional quadrotors solve this by vectoring thrust, but existing designs achieve it through heavy, mechanically complex, or structurally invasive changes to the airframe. This paper's goal is a modular, lightweight alternative: a gimballed rotor platform that bolts onto a standard quadrotor arm with only one added servo per rotor, plus a new control allocation scheme in PX4 Autopilot to command both rotors and servos.

Key Contributions

1. A gimballed rotor mechanism that houses a servo motor directly in the rotor base clamp, allowing each rotor to tilt independently without modifying the vehicle's central frame. Only four servo motors are added to a conventional quadrotor.
2. A complete flight dynamics model of the gimballed-rotor quadrotor, including a systematic derivation of propulsive forces and moments that the authors state can be extended to vehicle configurations with an arbitrary number of gimballed rotors.
3. A new control allocation scheme implemented in PX4 Autopilot, via new classes derived from ActuatorEffectiveness and ControlAllocation, replacing the standard body-frame thrust-vector-to-attitude mapping with a pseudo-inverse allocation that outputs both rotor speeds and tilt angles.
4. A fabricated prototype and successful flight tests, including a level-flight trajectory tracking experiment demonstrating decoupled position and attitude control compared to the same quadrotor with the tilting mechanism disabled.

Main Findings

• Component-level requirements were met or exceeded. The design targets were a 13 N maximum thrust withstand (CLR-1), a ±20° tilt range driven by a 1g lateral acceleration requirement and a 3:1 thrust-to-weight ratio (CLR-2), a 0.3 Nm minimum torque based on a 5 N average disturbance force and a 6 cm moment arm (CLR-3), and a tilting speed greater than 7.85 rad/s (CLR-4, taken from prior work). The built mechanism achieves a ±40° range of motion and a lateral acceleration of 10 m/s², exceeding CLR-2.
• Minimal size and mass penalty. The mechanism is 45 mm wide versus 35 mm for the original rotor platform, an increase the authors describe as modest and intended to limit aerodynamic interference and thrust loss.
• Direct-drive actuation removes gearing. The servo connects directly to the rotor base, eliminating the gear systems and lever arms found in other omnidirectional designs and reducing points of failure.
• Only partial SE(3) coverage. Because the mechanism has a ±40° range of motion, the design does not achieve the full SE(3) flight envelope of comparable vehicles; it is rated 1/5 on frame modification complexity with 4 motors + 4 servos, versus 2/5 for Voliro (12 motors), 4/5 for ODAR (16 motors), 3/5 for Omnicopter (8 motors), and 5/5 for FAM (12 motors), all of which reach entire SE(3).
• Level attitude was maintained during flight. In the flight test with gimballed rotors enabled, the vehicle held a level attitude throughout the mission; abrupt roll and pitch increases occurred only at waypoints where the vehicle transitioned perpendicular to its path.
• Yaw oscillations were observed. The omnidirectional quadrotor showed yaw oscillation, attributed to aerodynamic interference (assumed minimal in the model) and to a bias introduced when the vehicle was positioned before takeoff.
• Conventional baseline showed uneven motor loading. With the tilting mechanism disabled, counterclockwise propellers ran at an average normalized speed of 0.7 versus approximately 0.45 for clockwise propellers, while yaw stayed close to zero, indicating a natural yaw tendency from uneven weight distribution caused by hardware placement.
• Prototype total mass was 2045 g. The bill of materials lists the gimballed rotors at 4 × 55 g (220 g total), the structural frame at 610 g, power at 500 g, the flight computer at 175 g, propulsion at 4 × 75 g (300 g), flight control at 40 g, communications at 30 g, and miscellaneous components at 170 g.

Methodology in Plain English

The authors started from an off-the-shelf Holybro X500 quadrotor and asked what minimal change would let it move in any direction without tilting its body. Their answer was to replace each original rotor platform with a two-part 3D-printed assembly: a base clamp that holds a small servo motor and provides rails and tracks, and a rotating platform that carries the rotor and attaches to the servo arm. The servo tilts the rotor about an axis aligned with the vehicle's center of mass.

They set explicit design targets for thrust, tilt angle, torque, and tilt speed, then chose materials and hardware to satisfy them. The structural parts are 3D printed from carbon-fiber-reinforced high-temperature polyamide, rated for normal stresses up to 125 MPa under bending loads, with a density of approximately 1.06 g/cm³ and a heat deflection temperature of approximately 194°C. The actuator is a SAVOX SV-1232MG Micro Servo, weighing 23 g and measuring 23 mm × 12 mm × 27.3 mm, delivering 5.0 kg-cm of torque and 20.94 rad/s of speed at its maximum operating voltage of 7.4 V.

For control, they wrote the standard rigid-body equations of motion for an X-configuration vehicle, then derived how each rotor's thrust and torque map into body-frame forces and moments as a function of tilt angle. They assembled this into a 6×12 control effectiveness matrix and inverted it with a pseudo-inverse, so that a desired force-and-torque vector could be converted into individual rotor thrusts, rotor speeds, and tilt angles. They implemented this inside PX4 by modifying the position controller to always request a level attitude setpoint and overriding the function that normally converts a thrust vector into an attitude command, and by writing new control allocation classes.

Flight testing was conducted indoors with two autonomous missions following four waypoints in a square pattern: one with the tilting mechanism disabled as a conventional-quadrotor baseline, and one with it enabled. Position feedback came from an OptiTrack motion capture system and attitude feedback from the Pixhawk 6C's built-in inertial measurement unit.

Why This Matters

Impact on research. Most omnidirectional multirotor designs achieve full SE(3) flight at the cost of many actuators and substantial airframe surgery. This paper shows a middle path: partial SE(3) capability from four motors plus four servos on an otherwise unmodified frame, implemented on a widely used open-source autopilot stack. It also contributes a reusable modeling and allocation formulation that the authors note extends to other rotor counts and configurations.

Real-world applications.

• Inspection of surfaces and structures, where a vehicle can hold a level attitude while translating sideways or hovering at an angle.
• Indoor navigation and flight in confined spaces, where the ability to move without reorienting reduces the swept volume of the vehicle.
• Aerial manipulation, where keeping the body level simplifies interaction with objects and payloads.
• Operation in environments with large external disturbances, where the ability to decouple position from attitude improves disturbance rejection and precision.

Industry relevance. The design deliberately relies on off-the-shelf components, PX4 Autopilot, and ROS, and on standard fabrication tools such as a screwdriver and pliers for assembly and disassembly. The modular structure means a crash or failure of one gimbal does not render the whole system inoperable, unlike designs that require extensive central-body modification. These choices lower the barrier for developers and manufacturers to replicate, repair, and adapt the platform.

Future Directions

• Extending to dual-axis gimballed rotors, which the authors name explicitly as future work, would broaden the achievable thrust-vector directions beyond the current single tilt axis.
• Implementing a dedicated controller instead of PX4's cascaded position-then-attitude architecture, which the authors identify as a limitation of the current implementation.
• Addressing aerodynamic interference, since the flight tests showed yaw oscillations and the modeling assumption that rotor thrusts act independently without mutual interference holds only for small tilting angles.
• Resolving the gap between modeled and observed behavior, particularly the yaw oscillation, the initial yaw bias from pre-takeoff positioning, and the natural yaw tendency from uneven hardware weight distribution observed in the conventional baseline.

Target Audience

This paper is most useful to aerial robotics researchers and graduate students working on omnidirectional or overactuated multirotors, and to engineers who want a mechanically simple, modular hardware design they can build with standard 3D printing and off-the-shelf parts. Developers already working with PX4 Autopilot will find the control allocation modifications directly relevant, while readers focused on mechanisms alone can skip the modeling and control sections without losing the design rationale.

Authors’ abstract

This paper presents the design of a gimballed rotor mechanism as a modular and efficient solution for constructing omnidirectional quadrotors. Unlike conventional quadrotors, which are underactuated, this class of quadrotors achieves full actuation, enabling independent motion in all six degrees of freedom. While existing omnidirectional quadrotor designs often require significant structural modifications, the proposed gimballed rotor system maintains a lightweight and easy-to-integrate design by incorporating servo motors within the rotor platforms, allowing independent tilting of each rotor without major alterations to the central structure of a quadrotor. To accommodate this unconventional design, we develop a new control allocation scheme in PX4 Autopilot and present successful flight tests, validating the effectiveness of the proposed approach.