[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Next Article in Journal
Foggy Drone Teacher: Domain Adaptive Drone Detection Under Foggy Conditions
Previous Article in Journal
Research on Super-Twisted Sliding Mode Anti-Disturbance of UAV-Mounted Optoelectronic Platform Based on Predictive Adaptive Law
Previous Article in Special Issue
Dynamic Inversion-Enhanced U-Control of Quadrotor Trajectory Tracking
You seem to have javascript disabled. Please note that many of the page functionalities won't work as expected without javascript enabled.
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Design and Control Strategies of Multirotors with Horizontal Thrust-Vectored Propellers

by
Ricardo Rosales Martinez
*,
Hannibal Paul
and
Kazuhiro Shimonomura
Department of Robotics, Graduate School of Science and Engineering, Ritsumeikan University, Kusatsu 525-8577, Japan
*
Author to whom correspondence should be addressed.
Drones 2025, 9(2), 145; https://doi.org/10.3390/drones9020145
Submission received: 30 December 2024 / Revised: 7 February 2025 / Accepted: 14 February 2025 / Published: 16 February 2025
(This article belongs to the Special Issue Dynamics Modeling and Conceptual Design of UAVs)
Figure 1
<p>Multirotors equipped with horizontal thrust-vectoring systems. The red arrows in the image show the positions of the horizontal thrusters: (<b>a</b>) Performing torsional work using the with 3 vectored thrusters. (<b>b</b>) Conducting high-pressure washing using with 3 fixed thrusters. (<b>c</b>) Establishing wall contact using dual thrusters (<b>d</b>) Enabling planar motion with a single thruster.</p> ">
Figure 2
<p>Topologies of multirotors with thrusters: (<b>a</b>) Romero et al. (2007) [<a href="#B12-drones-09-00145" class="html-bibr">12</a>], (<b>b</b>) Albers et al. (2010) [<a href="#B13-drones-09-00145" class="html-bibr">13</a>], (<b>c</b>) Imamura et al. (2016) [<a href="#B15-drones-09-00145" class="html-bibr">15</a>], (<b>d</b>) González-deSantos et al. (2020) [<a href="#B14-drones-09-00145" class="html-bibr">14</a>], (<b>e</b>) Orjales et al. (2021) [<a href="#B19-drones-09-00145" class="html-bibr">19</a>], (<b>f</b>) Miyazaki et al. (2022) [<a href="#B16-drones-09-00145" class="html-bibr">16</a>], (<b>g</b>) Martinez et al. (2023) [<a href="#B17-drones-09-00145" class="html-bibr">17</a>], (<b>h</b>) Martinez et al. (2024) [<a href="#B18-drones-09-00145" class="html-bibr">18</a>], and (<b>i</b>) Martinez et al. (2024) [<a href="#B18-drones-09-00145" class="html-bibr">18</a>].</p> ">
Figure 3
<p>Flight controller block diagram.</p> ">
Figure 4
<p>Generalized topology for actuators with horizontal thruster support.</p> ">
Figure 5
<p>Block diagram illustrating the communication flow within the developed system.</p> ">
Figure 6
<p>Setup for torque evaluation (<b>a</b>). Dual-thruster configuration. (<b>a.1</b>) Moment evaluation with dual-thruster (<b>b</b>). Three-thruster configuration (<b>b.1</b>) Moment evaluation with 3 thrusters.</p> ">
Figure 7
<p>Moment evaluation results.</p> ">
Figure 8
<p>Thrusters at a horizontal offset from the center of mass.</p> ">
Figure 9
<p>Trajectory and angle setpoints during flight: (<b>a</b>) forward motion (+x), (<b>b</b>) backward motion (−x), (<b>c</b>) lateral right motion (+y), and (<b>d</b>) lateral left motion (−y).</p> ">
Figure 10
<p>Horizontal thrust-vectored thruster flight results. (<b>a</b>) forward motion (+x), (<b>b</b>) backward motion (−x), (<b>c</b>) lateral right motion (+y), and (<b>d</b>) lateral left motion (−y).</p> ">
Figure 11
<p>Thrusters positioned at vertical offsets from the center of mass of the UAV.</p> ">
Figure 12
<p>Thrusters positioned at negative offset from the center of mass of the UAV (0 mm to −60 mm).</p> ">
Figure 13
<p>Thrusters positioned at negative offset from the center of mass of the UAV (−80 mm to −100 mm).</p> ">
Figure 14
<p>Thrusters positioned at positive offset from the center of mass of the UAV (0 mm to 60 mm).</p> ">
Figure 15
<p>Thrusters positioned at positive offset from the center of mass of the UAV (80 mm to 100 mm).</p> ">
Review Reports Versions Notes

Abstract

:
With the growing adoption of Unmanned Aerial Vehicles (UAVs) in industrial and commercial sectors, the limitations of traditional under-actuated multirotors are becoming increasingly evident, particularly in manipulation tasks. Limited control over the thrust vector direction of the propellers, coupled with its interdependence on the vehicle’s roll, pitch, and yaw moments, significantly restricts manipulation capabilities. To address these challenges, this work presents a control framework for multirotor UAVs equipped with thrust-vectoring components, enabling enhanced control over the direction of lateral forces. The framework supports various actuator configurations by integrating fixed vertical propellers with horizontally mounted thrust-vectoring components. It is capable of handling horizontal thruster setups that generate forces in all directions along the x- and y-axes. Alternatively, it accommodates constrained configurations where the vehicle is limited to generating force in a single direction along either the x- or y-axis. The supported UAVs can follow transmitter commands, setpoints, or predefined trajectories, while the flight controller autonomously manages the propellers and thrusters to achieve the desired motion. Moment evaluations were conducted to assess the torsional capabilities of the vehicles by varying the angles of the thrusters during torsional tasks. The results demonstrate comparable torsional magnitudes to previously studied thrust-vectoring UAVs. Simulations with vehicles of varying inertia and dimensions showed that, even with large horizontal thruster offsets, the vehicles followed desired trajectories while maintaining stable horizontal orientation and smaller attitude variations compared to normal flight. Similar performance was observed with positive and negative vertical offsets, demonstrating the framework’s tolerance for thrusters outside the horizontal plane.

1. Introduction

Unmanned Aerial Vehicles (UAVs) can be categorized based on the nature of the forces and moments generated by their propulsion systems and actuator configurations [1,2]. The most common type is under-actuated vehicles [3,4,5,6], which feature vertically aligned propellers. These UAVs generate forces and moments by tilting their entire structure to redirect the thrust vector, coupling attitude control with thrust orientation. While under-actuated UAVs are relatively simple to construct and operate with well-established control systems, their design limits the range of forces they can apply, thereby restricting operational flexibility.
Despite these constraints, under-actuated UAVs have gained significant interest in non-destructive testing (NDT) applications [7], including non-contact tasks such as visual inspections of power lines and infrastructure [8]. However, for contact-based NDT tasks requiring direct force application, more specialized configurations have been explored. For instance, ref. [3] demonstrated an under-actuated multirotor equipped with an articulated arm mounted on top of its frame, enabling effective force application during bridge slab inspections. Similarly, refs. [4,5] showcased controlled lateral force application for ultrasonic inspection tasks. Additionally, ref. [6] introduced a platform combining a quadrotor with active suction cups for wall perching and a tiltable tool table, facilitating drilling operations on vertical surfaces.
Fully actuated UAVs [9,10,11] employ non-collinear propeller configurations, effectively decoupling attitude control from force generation. This design allows them to exert greater forces and perform more intricate manipulation tasks, albeit at the cost of increased construction and control complexity. In [9], a fully actuated UAV with tilted coplanar propellers is equipped with an eddy current (EC) sensor for assessing the state of welds. Similarly, ref. [10] introduces a UAV with tilting propellers to interact with curved surfaces, while ref. [11] extends this concept with a rolling sensor to evaluate concrete corrosion along structural surfaces.
For contact inspection tasks, such as those performed on concrete walls or bridges [4,6,11], the force requirements often exceed the capabilities of under-actuated UAVs. Addressing these requirements does not necessarily demand the added complexity of fully actuated systems. Practical solutions include equipping under-actuated UAVs with additional horizontal propellers, referred to as thrusters in this text, which enable the generation of controlled lateral forces [12,13,14,15,16,17,18,19]. This enhancement improves their ability to interact effectively with target surfaces while allowing planar motion and expanding their range of movement capabilities. Additionally, horizontal linear motion can also be achieved by guiding the thrust of multirotors through the use of inclined sub-rotor control surfaces, which have proven effective in addressing similar challenges [20].
Building on this premise, our research introduces a generalized control framework that integrates horizontal thrust components into under-actuated UAVs, expanding their manipulation and mobility capabilities. Our framework supports various thruster configurations, including tiltable thrusters, and introduces new flight modes to accommodate diverse task requirements. This work builds upon our previous research [16,17], where electric ducted fans (EDFs) were integrated to augment both the translational and lateral force capabilities of UAVs [16], as depicted in Figure 1a. Additionally, in [17], thrust-vectoring thrusters were employed to generate significant torsional moments, as shown in Figure 1b. The vehicles shown in Figure 1c,d combine the flight capabilities of dual- and single-thrust topologies [18]. These configurations enable both autonomous and manual flight modes, integrating the thrusters’ capabilities and facilitating contact interactions with surfaces. To realize these advancements, the proposed control framework has been implemented within the open-source PX4 Autopilot Firmware [21]. The source code developed for this work, including the simulated models, controller gains, and detailed instructions for replication, is available on GitHub https://github.com/rjros/Horizontal_thrusters (accessed on 13 February 2025). This implementation includes model-based controllers that improve stability for vehicles with non-uniform inertias and mass distributions, as well as support for tiltable horizontal thrusters. The framework unifies the control of various configurations, providing a foundation for future UAV applications in contact inspection. The main contributions of this paper are summarized as follows:
  • Development of a generalized, open-source control framework for multirotors with tiltable and fixed horizontal thrusters.
  • Inclusion of model-based controllers to enhance stability for vehicles with uneven inertia and mass distributions.
  • Comparative analysis of different thruster configurations and the effects of their placement on system performance.

2. Design Topologies for Multirotors with Horizontal Thrusters

The following section examines the characteristics and limitations of multirotor implementations with additional thrusters, as explored in previous studies [12,13,14,15,16,17,18,19]. Given that these vehicles share certain design characteristics with fully actuated systems, the scope of the analysis is limited to vehicles where only the secondary thrusters generate lateral forces. The general dimensions of these vehicles are provided in Figure 2, along with a detailed illustration of their configuration.
In addition to the forces generated by different actuator configurations, the arrangement of actuators influences the available flight modes and the level of autonomy during flight. Autonomous flight refers to vehicles that can maintain their position in space and follow setpoints or trajectories, even in the presence of disturbances. Semi-autonomous flight, on the other hand, refers to vehicles that have partial autonomy, controlling certain aspects of flight—such as attitude or movement in specific directions—while other motions still require manual control via a transmitter.
The main properties of each topology, including the number of actuators, types of thrusters, thruster forces, and UAV and thruster control type, are summarized in Table 1.
The control strategies for these systems can be broadly classified into two categories: multirotors with independent control architectures and multirotors with integrated control architectures.
The aerial vehicles shown in Figure 2a–g describe independent control architectures, where the additional thrusters function as modular add-on components, as discussed in [12,13,14,15,16,17,19]. In these designs, the control of the additional thrusters operates independently from the primary flight controller unit (FCU). The FCU uses the main vertical propellers control the altitude and maintain the desired attitude. The additional horizontal thrusters are controlled by a separate PC, with dedicate position and velocity controllers. Autonomous flight when utilizing the horizontal thrusters is restricted to configurations capable of generating forces along both the x- and y-axes. To illustrate this capability, the following vehicles showcase different morphologies designed to produce lateral forces in the x y -plane.
The system shown in Figure 2a depicts an eight-rotor UAV equipped with two separate microprocessors: one dedicated to controlling the vertical propellers and the other to the horizontal propellers. Using measurements from the IMU and GPS, the four vertical propellers manage the vehicle’s altitude and attitude. Meanwhile, the four horizontal propellers utilize position data from the GPS or an optical flow sensor to control lateral movement.
Figure 2e shows an octocopter equipped with four bidirectional horizontal propellers, designed for tasks involving vertical walls where the UAV must either fly in close proximity or make direct contact. It is equipped with ultra-wideband sensors and LIDAR-based range sensors to estimate its position relative to the wall. The Near Wall Positioning System (NWPS) enables a controlled approach using the horizontal thrusters. The total weight of the payload related to the thrusters and its components is less than 200 g.
The configuration in Figure 2f describes a hexacopter platform, equipped with three EDFs positioned at 120° separation from each other. This setup provides the minimum number of thrusters required for planar movement in the x y -plane. The combined weight of the three EDFs and their controller is 0.83 kg. Flight control is managed by a DJI N3 flight controller paired with an onboard PC. Serial communication enables the onboard PC to interface with the flight controller, allowing it to receive IMU data and RC channel input signals, as well as send attitude commands to the FCU. During planar flight, attitude setpoints are sent to the FCU to maintain a horizontal orientation and control the UAV’s altitude. Positioning in the x y -plane is regulated by the EDFs, utilizing feedback from a stereo tracking camera.
An extension of Figure 2f’s work is presented in Figure 2g where the UAV is equipped with additional thrust-vectoring components, which weigh 1.01 kg. This allows the system to change between the previous planar mode and a high-torque mode. While the main propellers of a multirotor generate relatively low moment in the yaw direction, by controlling the thrust direction of the horizontal thrusters, the UAV can produce significantly higher torsional moments. After landing on a vertically oriented industrial valve, the vertical propellers can be disarmed, and the vehicle transitions to high-torque mode. This increased moment is sufficient to turn the valve from its shut state.
In applications where generating forces in a single direction is the focus, such as in the inspection of vertical surfaces, fewer thrusters are needed. However, this often limits the system’s control to semi-autonomous or manual operation when the thrusters are engaged. The following vehicles illustrate different morphologies designed to produce limited lateral forces.
Figure 2b shows a quadrotor equipped with a single horizontal thruster, which is designed for controlled approach and force application on a vertical wall. Ultrasonic sensors calculate the distance and yaw angle of the UAV relative to the wall. As the UAV approaches the wall, the FCU manages yaw and altitude control, while a separate microcontroller regulates the horizontal thruster. The total weight of the system is 1.4 kg, including the payload and components.
The research presented in Figure 2d enables the semi-autonomous control of a quadrotor equipped with two horizontally aligned propellers. The add-on structure, which houses the distance sensors, horizontal thrusters, and manipulator, is referred to as an active payload. Similar to the setup in Figure 2b, feedback from the distance sensors is used to control the thrusters and the yaw angle when approaching the surface of interest. A separate controller inside the active payload sends attitude commands to the quadrotor’s FCU to maintain a horizontal orientation and control the yaw.
Figure 2c describes a quadrotor equipped with two EDFs mounted on a two-axis gimbal mechanism, weighing a total of 2.59 kg. The thrust-vectoring capabilities of this configuration allow for the generation of lateral forces by adjusting the angles of the EDFs. In this setup, the FCU handles attitude and altitude stabilization, while the operator controls horizontal motion directly through remote commands.
From the presented examples and configurations, the following key points emerge:
  • Separating the control logic between the thrusters and multirotor propellers simplifies overall system control, especially in configurations that generate forces along both the x- and y-axes. However, autonomous flight in more restricted or limited thruster configurations, such as those in vehicles (b), (c), and (d), requires a greater degree of control over the FCU.
  • The previous examples demonstrate design flexibility in terms on the supported UAV sizes, the number of propellers and thrusters, as well as their configurations. Additional thrust components can be strategically incorporated based on the specific requirements of the task. However, this variability also highlights a key weakness of this approach: the lack of generalization. Changes in actuator configurations require corresponding adjustments to the controllers, limiting the ability to apply a unified control strategy across different designs.
  • From the perspective of the flight controller, these additional actuators are indistinguishable from external disturbances, which can lead to unpredictable behaviors and may require failsafe design considerations.
In contrast, combined control architectures, such as the one described in [18], combines the control of the propellers and horizontal thrusters. This approach not only incorporates the flight capabilities of previous systems but also supports multiple actuator topologies. Multirotors within this architecture can be controlled both autonomously and manually, with thruster configurations that allow forces to be generated in all directions along the x- and y-axes. Additionally, more constrained configurations can be used, where forces are generated in only one direction along either the x- or y-axis. This, in turn, provides greater flexibility in designing aerial vehicles and determining the placement of the thrusters. The following vehicles illustrate the integrated control of propellers and thrusters.
Figure 2h depicts a quadrotor with two thrusters positioned along its x-axis but oriented in opposite directions. When following transmitter commands, setpoints, or trajectories, the FCU autonomously controls the propellers and thrusters to fulfill the desired motion. This vehicle is capable of planar motions along both directions of the x-axis. Lateral motions are accomplished by the control of its roll angle.
Figure 2i showcases a quadrotor equipped with an additional thruster aligned along its x-axis. This configuration allows the vehicle to execute planar motions exclusively when moving forward. To navigate backward, towards the negative direction of the x-axis, the system adjusts its pitch angle. Similarly, lateral motions are accomplished by controlling its roll angle.
To evaluate the integrated control architecture, the PX4 Autopilot [21], an open-source firmware commonly employed in both research and commercial vehicles, was used. Previous studies, such as those in [22,23], have successfully implemented enhancement to the PX4 firmware, enabling the control of various fully actuated vehicles. These studies also provided valuable insights into the modules of the PX4 firmware. By default, the flight controller supports a variety of vehicle types, such as multirotors, fixed-wing vehicles, and ground or underwater vehicles. Although the firmware allows customization of each actuator’s position and orientation, the control behavior and, ultimately, the flight type are determined by the selected vehicle classification. For vehicles classified as multirotors, the control behavior remains the same regardless of the orientation of the additional propellers, with the system controlling the attitude to tilt in the direction of the setpoints. A custom version of this firmware was developed, which integrated the control logic into the FCU. While in independent control architectures the controllers are managed by an onboard PC, this method centralizes control within the flight controller, making it easier to reproduce research, foster collaboration, and increase accessibility for potential users, particularly those with limited experience. This, in turn, frees the onboard PC to handle other processing tasks, such as state estimation, control of manipulators, and image processing. Additionally, in the event of communication interruptions with the onboard PC, the flight controller maintains control of the actuators responsible for flight.
However, this previous approach had certain limitations. The proposed controller only supported fixed actuator topologies, which meant that thrust-vectoring was not possible within the integrated architecture. Furthermore, transitions between the supported flight modes had to be managed manually.

3. Multirotors with Horizontal Thrust-Vectored Propellers

For the purpose of supporting the vehicles described in the previous section and providing a robust generalization of their models, a new control framework and custom firmware were developed. This controller integrates the unique configurations of these vehicles while extending their capabilities to support additional flight modes and actuator topologies. A diagram illustrating the developed control blocks within the PX4 architecture is presented in Figure 3. The current framework includes the Controller Mode block, which allows users to select between the PX4 Multicopter (MC) Controller and the controllers developed for thrust-vectored vehicles. In addition, a detailed explanation of the developed control blocks is provided in the following subsections.
The current stage of the firmware and the mathematical model presented in this work is based on several key assumptions and constraints, which are outlined as follows:
  • The firmware primarily supports vehicles with horizontal thrusters that operate within the horizontal plane. This implies that the resultant lateral thrust vectors must be perpendicular to the vertical propellers.
  • The supported thruster topologies are designed such that the sum of moments produced by the thrusters in the yaw direction is zero for during planar flight. This ensures that the vehicle maintains a stable orientation without any unwanted rotational forces in the horizontal plane. Furthermore, this assumption allows only the vertical propellers to control the yaw, ensuring that yaw moments are exclusively managed by these propellers.
  • Based on the physical limitations of the flight controller, the maximum number of propeller combinations, including both vertical and horizontal propellers, is limited to 16.
Based on these constraints, the proposed system can support most of the topologies described in Figure 2. However, support for the configuration shown in Figure 2c is limited, as the thrusters can generate forces in both the horizontal and vertical plane.
Throughout this paper, all vectors are expressed in the body frame of reference unless otherwise stated. Vectors not in the body frame are distinguished by specifying the frame of reference (e.g., BF for the body frame and WF for the inertial frame). The adopted frame convention is North-East-Down (NED) for the inertial frame F I , and Forward-Right-Down (FRD) for the body frame F B . In addition, x ^ B , y ^ B , and z ^ B represent the unit vectors aligned with the axes of the body frame. Table 2 refers to the notation used.

3.1. Control Allocation Module

Based on the geometric arrangement of the propellers and additional actuators, the Control Allocation block determines the control capabilities of the UAV. A control allocation matrix is generated containing the information of the position and orientation of all the actuators within the system. The default firmware employs a static allocation method for computing the angular velocities required to achieve the desired force and moment setpoints. The allocation is calculated before arming the vehicle, and remains constant during most of the flight. Given these constraints, a new vehicle type has been included, referred as thrust-vectoring vehicles in the control allocation block. This vehicle type supports the combined used of fixed vertical propellers and vectorable horizontal thrusters. Since the horizontal propellers can change their orientation, the control allocation matrix is dynamically allocated; in addition, the system makes a distinction between the vertical and horizontal propellers. Figure 4 illustrates the general case for the supported actuators. The vertical propellers, r n , include the different under-actuated multirotor configurations, with their thrust vectors facing the negative z-axis. The horizontal thrusters, r n * , can either have fixed orientations, where the thrust generates forces and moments based on their alignment with the center of mass (CoM), or they can be adjustable. In the latter case, the thrusters rotate about the z n * -axis, allowing the direction of the horizontal force to be modified to control the motion of the vehicle. The main propellers, referring to the fixed actuators, are capable of controlling the UAV as a normal multicopter vehicle, and do not need to be dynamically allocated during flight, since their position and orientation is fixed. Similar to the default multirotor setup, the allocation matrix A remains constant, representing a static allocation method, as shown below:
F m a i n M m a i n = A · λ
The generalized form of the allocation matrix A is shown in its expanded form as
f z τ x τ y τ z = μ μ μ μ y 1 μ y 2 μ y 3 μ y n μ x 1 μ x 2 μ x 3 μ x n μ γ 1 κ γ 2 κ γ 3 κ γ n κ ω 1 2 ω 2 2 ω 2 3 ω n 2
A secondary allocation matrix is reserved for the horizontal thrusters. When their tilt angle changes, the allocation matrix is updated to represent the new force distribution.
F v e c t M v e c t = A ( Θ ) · λ
where the structure of A ( Θ ) depends on the thrust vector angle θ n and must be dynamically allocated to be solved. During flight, as the thrusters change their orientation, the allocation matrix A ( Θ ) is updated. The generalized form of the dynamic allocation matrix A ( θ ) is shown in its expanded form as follows:
f x f y τ x τ y τ z = μ * C θ 1 μ * C θ 2 μ * C θ n μ * S θ 1 μ * S θ 2 μ * S θ n κ * C θ 1 κ * C θ 2 κ * C θ n κ * S θ 1 κ * S θ 2 κ * S θ n x 1 μ * S θ 1 + y 1 μ * C θ 1 x 2 μ * S θ 2 + y 2 μ * C θ 2 x n μ * S θ n + y n μ * C θ n ω 1 2 ω 2 2 ω n 2
From the control allocation matrices, the control parameter A x y is constructed, which contains information about the thrust capabilities of the current configuration. This vector in the form of [ + f x , f x , + f y , f y ] , provides the Position Controller’s Flight Mode block, shown in Figure 3, with the direction of the thrust forces that can be controlled by the vehicle. This enables the vehicle to automatically switch to the appropriate flight mode and distribute the desired forces among its actuators based on A x y and desired trajectory. Additionally, the operator has the option to manually switch between flight modes if needed.

3.2. Thrust-Vectoring Position Control in SE(3)

The Multicopter Position Control block computes the necessary thrust and attitude commands based on a provided set of initial position P s p and yaw ψ s p setpoints. P s p denotes either the operator’s transmitter commands, internal predefined trajectories, or trajectories specified by an on-board PC.
The PX4 firmware, in its default configuration, is designed to be largely model-invariant. It employs a standard control framework that can be applied to a wide range of vehicle configurations without requiring vehicle-specific parameters such as mass or inertia. Thrust calculations and the corresponding force magnitudes are represented in unitary values, enabling the flight controller to maintain its general applicability across standard multicopters. While this default configuration is suitable for many typical multicopter setups, it faces limitations when applied to custom vehicles with non-standard configurations or specialized flight modes. In such cases, key parameters like mass and inertia become critical for accurate control. These parameters are essential for advanced UAV control methods, such as geometric control methods [24], LQR-based controllers [25], and model predictive controllers [26,27]. To address this limitation, the developed framework includes a geometric position controller and allows the support of vehicles with non-uniform mass distribution and custom inertial characteristics.
The current system supports four primary flight modes: tilted flight mode, xy thruster mode, x thruster mode, and y thruster mode. The modes available for a UAV depend on the number and configuration of its actuators. Vehicles capable of generating planar forces along both the x- and y-axes can employ all the control strategies. On the other hand, UAVs with a more constrained actuator configuration, producing forces in limited directions, are designed to transition specifically between tilted mode and the x or y thruster mode.
  • Tilted Flight Mode: This mode refers to the normal flight mode where the vehicle controls its attitude, i.e., roll, pitch, and yaw, to fulfill the desired position, velocity, and acceleration setpoints. The position control of the UAV platform on the SE(3) controller is based on the geometric controller described in [24]. The tracking errors for the position and velocity are expressed as follows:
    e p = R y a w T ( p W p s p W ) e v = R y a w T ( p W p s p W )
    where R y a w represents a pure rotation about the z-axis by the yaw angle setpoint ψ s p , as shown below:
    R y a w = cos ( ψ s p ) sin ( ψ s p ) 0 sin ( ψ s p ) cos ( ψ s p ) 0 0 0 1
    The integral terms can then be described as follows:
    e i = 0 t ( e v ( t ) + c 1 e p ( τ ) ) d τ
    where c 1 > 0 is a proportional gain that determines the contribution of the position error to the integral error. The integral error e i is then saturated by σ > 0 , which depends on the UAV’s vertical and horizontal thrust capabilities. A PID controller is applied to obtain the desired acceleration, v ˙ * . The gains of the controller depend on the flight mode, since the propellers and the horizontal thrusters have different dynamic effects. The gains K p , K i , and K d are expressed as diagonal matrices and follow the general form
    K = k x 0 0 0 k y 0 0 0 k z
    where the respective elements k x , k y , and k z , represent the proportional, integral, and derivative gains for the x-, y-, and z-axes.
    v ˙ * = K p e p K d e v K i e i + v ˙ s p
    F = m ( v ˙ * + g B )
    The force vector is then represented in the inertial frame by the transformation given by
    F W = R y a w ( F )
    The thrust to attitude block uses the derived force vector to calculate the attitude setpoint. The desired orientation is expressed by the unit vectors, x ^ B , y ^ B , and z ^ B , which are calculated as follows:
    z ^ B = F W F W
    x ^ p r o j = cos ψ s p sin ψ s p 0
    where x p r o j refers to a projection of the x-axis in the x y -plane.
    y ^ B = z ^ B × x ^ p r o j
    x ^ B = y ^ B × z ^ B
    The obtained attitude is then constrained by the maximum allowable tilt of the vehicle. The attitude setpoint can subsequently be expressed in quaternion form as q s p . Following this, the force setpoint F s p is expressed using the force vector calculated in Equation (10), as follows:
    f t h r u s t = F W · ( R e 3 )
    F sp = 0 0 f t h r u s t T
    where R is the rotation matrix defining the orientation of the body frame relative to the inertial frame, and e 3 = [ 0 , 0 , 1 ] T is the unit vector along the z-axis of the body frame.
  • XY Thruster Flight Mode: This flight mode is specifically designed for configurations with actuators capable of directly generating forces along the x- and y-axes of the vehicle. The roll and pitch angles are kept close to zero degrees, while the horizontal thrusters are used for planar motion. The control of the UAV platform with horizontal thrusters follows a similar formulation to that of the tilted flight mode. However, horizontal and vertical propellers have different effects on the UAV’s movement. Therefore, the matrix K includes gains associated with the respective thrusters, as shown below.
    K = k t h x 0 0 0 k t h y 0 0 0 k z
    where the respective elements k t h x and k t h y represent the proportional, integral, and derivative gains for thruster control, while k z corresponds to the gain used in the tilted flight mode. The individual force components ( F x , F y , F z ) are obtained from Equation (10), and the desired thrust setpoint F s p , is constructed as follows:
    F sp = F x F y F z T
    The attitude setpoint q s p is derived from the calculated unit vectors representing the desired orientation. For planar flight, the unit vector z B is aligned downward, parallel to the z-axis of the reference frame F I , shown as follows:
    z ^ B = 0 0 1 T
    The remaining unit vectors x ^ B and y ^ B , will be rotated in the xy-plane by the yaw angle setpoint ψ s p , as shown below:
    x ^ B = cos ψ s p sin ψ s p 0 T y ^ B = sin ψ s p cos ψ s p 0 T
  • X Thruster Flight Mode: This flight mode integrates control strategies from both tilted flight mode and xy thruster flight mode, enabling planar flight along the positive or negative x-axis, while controlling its roll and pitch for the remaining motions. Only the gains in the x-axis correspond to the thruster, while y and z use the gains of the tilt flight mode. Therefore, matrix K is then expressed as follows:
    K = k t h x 0 0 0 k y 0 0 0 k z
    As before, the individual force components ( F x , F y , F z ) are obtained from Equation (10). Since the vehicle uses both attitude control and horizontal thrusters to achieve motion, the desired thrust vector F s p must be separated into two components. The first component, F x , corresponds to the force distributed to the horizontal thrusters and is aligned with the x-direction. The second component, comprising the remaining force components ( F y , F z ) , are grouped into the vector F y z . This vector is used for altitude and attitude control and is defined as follows:
    F yz = 0 F y F z T
    With these force components, the desired thrust setpoint F s p is constructed as follows:
    f t h r u s t = F yz · ( R e 3 )
    F sp = F x 0 f t h r u s t T
    The attitude setpoint q s p is then derived from the force vector F y z , by calculating the unit vectors of the desired orientation. The body-frame z-axis unit vector, z ^ B , is computed as follows:
    F y z W = R y a w ( F yz )
    z ^ B = F y z W F y z W
    For planar motions in the x-axis of the body frame, the pitch angle is set to 0 deg. Consequently, x ^ B and y ^ B are defined as follows:
    x ^ B = cos ψ s p sin ψ s p 0
    y ^ B = z ^ B × x ^ B
  • Y Thruster Flight Mode: The thrust and attitude calculations follow the same logic as the x thruster flight mode, with the key difference being that the horizontal forces are constrained to the positive or negative y-axis. The gain matrix for this mode is given by
    K = k x 0 0 0 k t h y 0 0 0 k z
    The desired thrust vector, F s p , is decomposed into two components: the horizontal thrust, F y , and F x z , which is used for altitude and attitude control.
    F xz = F x 0 F z T
    f t h r u s t = F xz · ( R e 3 )
    F sp = 0 F y f t h r u s t T
    The attitude setpoint q s p is then derived from the force vector F x z . The unit vectors can then be calculated as follows:
    F x z W = R yaw ( F xz )
    For planar motions along the y-axis of the body frame, the roll angle is set to 0 deg. The unit vectors x ^ B and y ^ B are defined as follows:
    z ^ B = F x z W F x z W
    y ^ B = sin ψ s p cos ψ s p 0 T
    x ^ B = y ^ B × z ^ B

3.3. Attitude and Rate Control

Attitude Control Module: In the attitude controller, the angular acceleration, ω ˙ * , is calculated to derive the required moments. The system employs the default PX4 implementation of the attitude controller, which is based on the framework described in [28].
The attitude error, q e , is expressed using the quaternion representation of rotations.
q e = q 1 q s p
Here, q represents the current attitude of the vehicle, q 1 , its inverse quaternion, q s p is the desired attitude, and ⊗ denotes the quaternion multiplication operator. Once the attitude error is computed, the desired angular velocity, ω s p , is determined using the proportional control law:
ω s p = K q s g n ( q e , 0 ) q e , 1 : 3
K q R 3 refers to the attitude gain, sgn ( q e , 0 ) determines the rotation direction using the sign of the scalar part of the quaternion, and q e , 1 : 3 represents the axis of rotation from the vector part of the quaternion.
The body rate error, e ω , and the angular acceleration, e ω ˙ are expressed as follows:
e ω = ω ω s p
e ω ˙ = ω ˙ ω ˙ s p
The computed angular acceleration, ω ˙ * is then used to derive the desired moment M s p . Additionally, modifications were made to the rate controller to explicitly incorporate the system’s inertia, as expressed in the following equations:
ω ˙ * = K ω p e ω K ω i e ω K ω d e ω ˙
M s p = I B B ω ˙ * + ω × I B ω

4. Simulations

The control framework described earlier was assessed in the Gazebo simulation environment using the Software-In-The-Loop (SITL) capabilities of the PX4 firmware. This setup enables evaluation of the developed control framework under conditions similar to actual flight and facilitates the assessment of multiple configurations. The mass, inertia, and motor capabilities were selected to match the actual hardware used in our previous work [18].
The simulation environment runs on a Ubuntu 22.04 PC with a ROS2 Humble distribution, acting as the on-board PC. uXRCE-DDS is used for agent and client communication, providing control and navigation data to the flight controller in the SITL simulation. As illustrated in Figure 5, positioning data from the simulated environment are transmitted via ROS2 to the onboard PC, allowing the UAV to maintain its position and follow setpoints. For vehicles with tilting thrusters, the thrust vector publisher communicates orientation setpoints in the vector Θ s p . These angle setpoints are then used in the control allocation and mixer block, to control the position of the thrusters and update the allocation matrix.
The configurations evaluated consist of thrusters at various positions capable of changing their orientations. These horizontal thrusters are mounted on a common base vehicle, a Holybro X500 quadcopter in an x-configuration, with four main vertical propellers providing a maximum vertical thrust of 44.54 N. Each horizontal thruster, based on smaller propellers, is capable of generating a maximum thrust of 8.55 N. The vehicle and its actuators are controlled by the developed firmware. Through the open-source ground control station (GCS) software, users can change the configuration of the additional thrusters’ position and orientation. Custom parameters were created to tune the vehicle’s different controllers, limit horizontal velocities, and select flight modes for evaluation.

4.1. Moment Evaluation of Previous Configurations

The following simulations evaluate the torsional capabilities of different UAV configurations. The simulation setup is shown in Figure 6, where a force/torque sensor is placed between the UAV the cylindrical base. The vehicles evaluated included the dual-thruster system described in Table 3, and a three-thruster configuration model, previously tested in [17]. The thrusters’ placements on these vehicles allow for larger torque magnitudes to be generated when the thrusters’ orientations are changed. The maximum moment generated solely by the rotation of the vertical propellers occurs when only the propellers rotating in the same direction are active. In a quadrotor, this involves two propellers, with the resulting moment around the z-axis (yaw), induced by aerodynamic drag.
Figure 7 compares the moment generated by the vehicles. The vertical propellers generate the lowest torque values, with a magnitude of 0.35 N·m. In contrast, for the horizontal dual-thruster case, the torque values increase significantly, with the following corresponding magnitudes: 3.42 N·m for 0.20 m, 5.13 N·m for 0.30 m, 6.83 N·m for 0.40 m, and 8.55 N·m for 0.50 m. The highest torque magnitude was produced by the three-thruster case, reaching 13.60 N·m.

4.2. Thrust-Vectored Tilted Propeller Flight

The dual-thrust configuration, shown in Figure 8 with tilting horizontal thrusters, was selected as it requires the combination of all actuators to control the vehicle. The vehicles were designed with the thrust vector of the horizontal thrusters positioned in the same plane as the UAV’s center of mass (CoM). The thrusters were placed at a horizontal offset along the x-axis from the CoM to evaluate the effects of this placement during flight. The evaluated vehicles’ mass, inertia, and thruster placement are described in Table 3. The trajectory shown in Figure 9 evaluates the vehicle’s control performance along the x y -plane. The simulation assesses the UAV’s behavior in normal flight mode with the geometric controller, where the UAV tilts, and in thruster mode, where the thrusters are actively controlled. In the flight with the tilting thrusters, the vehicle follows the trajectory while adjusting the orientation of the horizontal thrusters ( θ 1 , θ 2 ) , to move along the plane when approaching the setpoint. The planar thrust is achieved by using horizontal thrusters in the direction of motion, and the remaining directions are controlled by the use of the main propellers. The vehicle can use its roll and pitch in combination with the horizontal thrusters to follow the desired trajectory.
Stages (a) and (b) represent forward and backward motion, respectively. In normal flight mode, the system tilts in the pitch direction to control its motion and speed. In thruster mode, the thruster angles are set to ( 0 ° , 0 ° ) . In this configuration, forces can be generated along the x-axis. The UAV utilizes the thrusters to move forward and backward while maintaining a pitch angle close to 0°. In both cases, the roll angle is adjusted for minor corrections to maintain the desired trajectory.
Stages (c) and (d) represent lateral motion. In normal flight mode, the roll angle is used for movement in the lateral directions, with the pitch angle providing minor adjustments. In thruster mode, the orientation of the thrusters is adjusted to generate thrust in the direction of motion.
For stage (c), the thruster angles are set to ( 90 ° , 90 ° ) , causing both thrusters to produce a horizontal force in the positive y-direction. In stage (d), the angles are set to ( 90 ° , 90 ° ) , generating a force vector in the negative y-direction. The roll angle remains close to 0° during these motions and it changes autonomously when the system needs to reduce speed and stop at the setpoint.
The vehicle’s position and attitude during the experiment are shown in Figure 10. Across all horizontal offsets, it can be observed that attitude changes are significantly larger during normal flight. In contrast, when using the thrusters, the attitude remains within smaller magnitudes, as the thrusters enable the system to move without tilting the frame for all motions. In thruster flight mode, during stages (c) and (d), the roll and pitch remain low, with values of approximately ± 0.1 ° and ± 3 ° , respectively. The roll exceeds 10 ° only when the vehicle comes to a stop, at 15 and 20 s.
As the horizontal offset increases and the mass distribution of the components changes, following the trajectory becomes increasingly challenging. At a 400 mm offset along both the x- and y-axes, the system is able to stop at the desired setpoint. However, during the speed reduction phases at second 6 in stage (a) and second 15.6 in stage (c), deviations of 0.14 m from the trajectory are observed in the normal flight. These are not present in the thrusters’ flight mode during stage (a), where the horizontal forces are utilized to reduce speed effectively. In stage (c), however, similar deviations occur as the thrusters are employed for motion but not for speed reduction.
For a 500 mm offset, small deviations are present in both flight modes. These deviations are likely due to the increased challenge in counteracting the effects of the larger horizontal offset and the altered mass distribution.

4.3. Vertical Offset Experiments

Optimal thruster placement is critical for achieving precise and efficient UAV control. Ideally, the thrusters should be aligned within the same plane as the UAV’s CoM. However, practical design constraints and task-specific requirements may present situations where this alignment is difficult to achieve. Therefore, the purpose of this experiment was to evaluate the effects of vertical offsets in the thruster placement. The vehicle is shown in Figure 11. Intuitively, this vertical offset induces a turning moment when the thrust vector acts on the vehicle. The evaluated vertical offsets included both positive and negative distances from the CoM. A positive offset refers to placing the thrusters above the CoM, while a negative offset refers to placing them below it. The vehicles, along with their mass and inertia, are shown in Table 4. Each vehicle’s controller was carefully tuned and evaluated. The simulation assessed the UAV’s behavior across two distinct flight modes: (1) a normal tilted flight mode utilizing a geometric controller and (2) a flight mode employing horizontal thrusters. The test trajectory involved forward and backward motion in both flight modes.
The vehicle’s position and attitude during the negative offset experiments are presented in Figure 12 and Figure 13. Similarly, Figure 14 and Figure 15 illustrate the position and orientation during the positive offset experiments.
In normal flight mode, an increase in vertical offset, whether positive or negative, results in a corresponding increase in the magnitude of the tilt along the pitch axis. For negative offsets, the variations remain relatively consistent throughout the flights. From −20 mm to −100 mm, the tilt is observed to stay within a range of approximately ± 15 ° . In contrast, positive offsets exhibit a gradual increase in the tilt magnitude with each additional offset. At 0 mm, the value is approximately ± 15 ° , while at 100 mm, it reaches ± 19 ° .
In thruster flight mode, deviations in attitude along the pitch axis are significantly reduced due to the system’s constraint of maintaining a level orientation. In the base case, where the thrusters are aligned with the CoM plane, these deviations are minimized. Throughout this mode, the variations remain below ± 5 ° , with the largest changes occurring at 80 mm and 100 mm offsets. Despite these variations, the tracking error remains unaffected. Even at the largest offset of 100 mm, the observed changes are only half of those recorded in the base case of normal flight mode.
Given that the motion is along the x-axis of the body frame, the roll angle exhibits similar values across all offsets and modes.

5. Discussion and Future Work

As shown in Figure 7, the torque magnitudes produced by conventional multirotor propellers significantly restrict their applications due to their limited torsional capabilities. These vehicles exhibit far lower rotational moments compared to thrust-vectored vehicles. Achieving comparable torque with conventional UAVs would necessitate larger sizes and increased weight, which could compromise their practicality and operational efficiency. The current system is limited to performing torsional tasks only after the system has landed. To enable torsional tasks during flight, the system must allocate the moments in the two allocation matrices based on the task requirements and the safe control of the UAV.
These compact versions are particularly promising, as they deliver high performance without the drawbacks associated with larger and heavier vehicles. While the three-thruster configuration previously evaluated in [17] showcased strong torsional capabilities, the more compact configurations simulated in this study are sufficient for many similar torsional tasks.
As shown in Figure 10, larger offsets yield higher torsional magnitudes, but introduce stability challenges and additional weight, which can adversely affect flight performance. For translational movements, large offsets are not beneficial. An adaptive design, such as telescopic joints, could allow the UAV to dynamically extend the thrusters only when higher torque is required, maintaining stability during regular operations.
The limitations of current horizontal thrust control with respect to vertical positioning are highlighted in Figure 11. While the control framework assumes thrusters operate in a common plane, the results indicate that certain deviations can be tolerated. This provides a foundation for further exploration into non-planar thruster arrangements, which could enhance torque and force generation at other directions and overall maneuverability. Moreover, even when the thrusters are not actively engaged, their added mass and the associated changes to the vehicle’s inertia must be carefully accounted for during the design and tuning phases. The results from the vertical offset tests in normal flight mode show that, although both positive and negative offsets are manageable, they do affect the vehicle’s attitude during flight. Specifically, negative offsets lead to smaller attitude variations compared to their positive counterparts. The placement of the thrusters and their associated mass causes the system to behave like a regular pendulum when the offsets are negative and like an inverted pendulum when the offsets are positive. However, further increasing the offsets and replacing the thrusters with more powerful ones will result in larger attitude variations. Incorporating this information into the framework could enhance flexibility in thruster positioning.
Future iterations will focus on validating the proposed framework through real-world experiments, examining different UAV topologies and actuator configurations. In addition to their impact on the forces generated, actuator configurations also affect power consumption. An energy consumption will be valuable for comparing under-actuated vehicles with thrusters to their fully actuated counterparts. Additionally, evaluations of the interaction effects between thrust-vectoring and vertical propellers, as well as ceiling and wall effects [29,30], will provide insights into optimal design criteria for selecting actuator topologies. These studies aim to refine the design, validate performance under diverse operating conditions, and expand the framework’s applicability to industrial and civil infrastructure inspection tasks.

6. Conclusions

In this research, a control and design framework has been introduced to enhance the manipulation capabilities of under-actuated multirotors by integrating additional thrust components. This approach aims to bridge the gap between the simplicity of under-actuated systems and the expanded capabilities of fully actuated vehicles, avoiding the increased complexity and cost associated with the latter. By strategically incorporating horizontal and tiltable thrusters, the framework enables a wide range of flight modes, providing the flexibility to adapt to diverse tasks and operational scenarios.
A key feature of this framework is its modularity, which allows it to accommodate different actuator configurations. This flexibility ensures that the system can be tailored to specific requirements, whether directional thrust or full planar force capabilities are needed. Additionally, the ability to transition between enhanced thruster flight modes and normal flight mode ensures efficient use of the thrusters, conserving energy and simplifying operations when additional thrust is not required.
The evaluated control framework was embedded directly into the flight controller, simplifying implementation and ensuring reproducibility. The developed firmware supports both autonomous and manual operation across all supported flight modes, with the ability to automatically switch modes according to actuator configurations. This integration reduces computational demand on the onboard PC, freeing it for advanced control algorithms. Embedding the framework also adds a critical safety feature: a backup controller within the FCU. This enables operators to safely land and disarm the system in the event of actuator failures, a particularly valuable capability when managing multiple actuators or tiltable propellers.
Through simulations, the performance of this framework has been evaluated with hybrid vehicle configurations. These tests demonstrate the framework’s ability to accommodate diverse designs while maintaining robust control and stability, even for systems with non-uniform inertia and mass distributions.

Author Contributions

Contributions of the authors are as follows: R.R.M. developed the formal analysis and software, performed experiments, analyzed the results, and wrote the manuscript. H.P. analyzed the design, performed experiments, and wrote the manuscript. K.S. provided design considerations, supervised, validated results, and wrote the manuscript. All authors have read and agreed to the published version of the manuscript.

Funding

This work was partially supported by JST CREST Grant Number JPMJCR22C1, Japan.

Data Availability Statement

No new data were created or analyzed in this study. Data sharing is not applicable to this article.

Conflicts of Interest

The authors declare no conflicts of interest.

Abbreviations

The following abbreviations are used in this manuscript:
CoMCenter of Mass
EDFElectric Ducted Fan
FCUFlight Controller Unit
FRDForward-Right-Down
IMUInertial Measurement Unit
NEDNorth-East-Down
NDTNon-destructive Testing
SITLSoftware-In-The-Loop
UAVUnmanned Aerial Vehicle

References

  1. Hamandi, M.; Usai, F.; Sablé, Q.; Staub, N.; Tognon, M.; Franchi, A. Design of multirotor aerial vehicles: A taxonomy based on input allocation. Int. J. Robot. Res. 2021, 40, 1015–1044. [Google Scholar] [CrossRef]
  2. Bodie, K.; Brunner, M.; Allenspach, M. Omnidirectional Tilt-Rotor Flying Robots for Aerial Physical Interaction: Modelling, Control, Design and Experiments; Springer Nature: Berlin/Heidelberg, Germany, 2023; Volume 157. [Google Scholar]
  3. Ikeda, T.; Yasui, S.; Minamiyama, S.; Ohara, K.; Ashizawa, S.; Ichikawa, A.; Okino, A.; Oomichi, T.; Fukuda, T. Stable impact and contact force control by UAV for inspection of floor slab of bridge. Adv. Robot. 2018, 32, 1061–1076. [Google Scholar] [CrossRef]
  4. Zhang, D.; Watson, R.; MacLeod, C.; Dobie, G.; Galbraith, W.; Pierce, G. Implementation and evaluation of an autonomous airborne ultrasound inspection system. Nondestruct. Test. Eval. 2022, 37, 1–21. [Google Scholar] [CrossRef]
  5. González-deSantos, L.; Martínez-Sánchez, J.; González-Jorge, H.; Navarro-Medina, F.; Arias, P. UAV payload with collision mitigation for contact inspection. Autom. Constr. 2020, 115, 103200. [Google Scholar] [CrossRef]
  6. Dautzenberg, R.; Küster, T.; Mathis, T.; Roth, Y.; Steinauer, C.; Käppeli, G.; Santen, J.; Arranhado, A.; Biffar, F.; Kötter, T.; et al. A Perching and Tilting Aerial Robot for Precise and Versatile Power Tool Work on Vertical Walls. In Proceedings of the 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Detroit, MI, USA, 1–5 October 2023; pp. 1094–1101. [Google Scholar] [CrossRef]
  7. Nooralishahi, P.; Ibarra-Castanedo, C.; Deane, S.; López, F.; Pant, S.; Genest, M.; Avdelidis, N.P.; Maldague, X.P.V. Drone-Based Non-Destructive Inspection of Industrial Sites: A Review and Case Studies. Drones 2021, 5, 106. [Google Scholar] [CrossRef]
  8. Fang, Z.; Savkin, A.V. Strategies for Optimized UAV Surveillance in Various Tasks and Scenarios: A Review. Drones 2024, 8, 193. [Google Scholar] [CrossRef]
  9. Tognon, M.; Chávez, H.A.T.; Gasparin, E.; Sablé, Q.; Bicego, D.; Mallet, A.; Lany, M.; Santi, G.; Revaz, B.; Cortés, J.; et al. A Truly-Redundant Aerial Manipulator System With Application to Push-and-Slide Inspection in Industrial Plants. IEEE Robot. Autom. Lett. 2019, 4, 1846–1851. [Google Scholar] [CrossRef]
  10. Bodie, K.; Brunner, M.; Pantic, M.; Walser, S.; Pfändler, P.; Angst, U.; Siegwart, R.; Nieto, J.I. An Omnidirectional Aerial Manipulation Platform for Contact-Based Inspection. arXiv 2019, arXiv:1905.03502. [Google Scholar]
  11. Pfändler, P.; Bodie, K.; Crotta, G.; Pantic, M.; Siegwart, R.; Angst, U. Non-destructive corrosion inspection of reinforced concrete structures using an autonomous flying robot. Autom. Constr. 2024, 158, 105241. [Google Scholar] [CrossRef]
  12. Romero, H.; Salazar, S.; Sanchez, A.; Lozano, R. A new UAV configuration having eight rotors: Dynamical model and real-time control. In Proceedings of the 2007 46th IEEE Conference on Decision and Control, New Orleans, LA, USA, 12–14 December 2007; pp. 6418–6423. [Google Scholar] [CrossRef]
  13. Albers, A.; Trautmann, S.; Howard, T.; Nguyen, T.A.; Frietsch, M.; Sauter, C. Semi-autonomous flying robot for physical interaction with environment. In Proceedings of the 2010 IEEE Conference on Robotics, Automation and Mechatronics, Singapore, 28–30 June 2010; pp. 441–446. [Google Scholar] [CrossRef]
  14. González-deSantos, L.M.; Martínez-Sánchez, J.; González-Jorge, H.; Arias, P. Active UAV payload based on horizontal propellers for contact inspections tasks. Measurement 2020, 165, 108106. [Google Scholar] [CrossRef]
  15. Imamura, A.; Miwa, M.; Hino, J. Flight characteristics of quad rotor helicopter with thrust vectoring equipment. J. Robot. Mechatronics 2016, 28, 334–342. [Google Scholar] [CrossRef]
  16. Miyazaki, R.; Paul, H.; Kominami, T.; Martinez, R.R.; Shimonomura, K. Flying Washer: Development of High-Pressure Washing Aerial Robot Employing Multirotor Platform with Add-On Thrusters. Drones 2022, 6, 286. [Google Scholar] [CrossRef]
  17. Martinez, R.R.; Paul, H.; Shimonomura, K. Aerial Torsional Work Utilizing a Multirotor UAV with Add-on Thrust Vectoring Device. Drones 2023, 7, 551. [Google Scholar] [CrossRef]
  18. Martinez, R.R.; Paul, H.; Shimonomura, K. Control Framework for Multirotors with Additional Horizontal Thrusters. In Proceedings of the 2024 IEEE International Conference on Advanced Intelligent Mechatronics (AIM), Boston, MA, USA, 15–18 July 2024; pp. 106–113. [Google Scholar] [CrossRef]
  19. Orjales, F.; Losada-Pita, J.; Paz-Lopez, A.; Deibe, Á. Towards precise positioning and movement of UAVs for near-wall tasks in GNSS-denied environments. Sensors 2021, 21, 2194. [Google Scholar] [CrossRef] [PubMed]
  20. Cetinsoy, E. Design and simulation of a holonomic quadrotor UAV with sub-rotor control surfaces. In Proceedings of the 2012 IEEE International Conference on Robotics and Biomimetics (ROBIO), Guangzhou, China, 11–14 December 2012; pp. 1164–1169. [Google Scholar] [CrossRef]
  21. Meier, L.; Honegger, D.; Pollefeys, M. PX4: A node-based multithreaded open source robotics framework for deeply embedded platforms. In Proceedings of the 2015 IEEE International Conference on Robotics and Automation (ICRA), Seattle, WA, USA, 26–30 May 2015; pp. 6235–6240. [Google Scholar] [CrossRef]
  22. Keipour, A.; Mousaei, M.; Ashley, A.; Scherer, S. Integration of Fully-Actuated Multirotors into Real-World Applications. arXiv 2020, arXiv:2011.06666. [Google Scholar]
  23. D’Angelo, S.; Pagano, F.; Longobardi, F.; Ruggiero, F.; Lippiello, V. Efficient Development of Model-Based Controllers in PX4 Firmware: A Template-Based Customization Approach. In Proceedings of the 2024 International Conference on Unmanned Aircraft Systems (ICUAS), Chania, Greece, 4–7 June 2024; pp. 1155–1162. [Google Scholar] [CrossRef]
  24. Lee, T.; Leok, M.; McClamroch, N.H. Geometric tracking control of a quadrotor UAV on SE(3). In Proceedings of the 49th IEEE Conference on Decision and Control (CDC), Atlanta, GA, USA, 15–17 December 2010; pp. 5420–5425. [Google Scholar] [CrossRef]
  25. Allenspach, M.; Bodie, K.; Brunner, M.; Rinsoz, L.; Taylor, Z.; Kamel, M.; Siegwart, R.; Nieto, J. Design and optimal control of a tiltrotor micro-aerial vehicle for efficient omnidirectional flight. Int. J. Robot. Res. 2020, 39, 1305–1325. [Google Scholar] [CrossRef]
  26. Baca, T.; Hert, D.; Loianno, G.; Saska, M.; Kumar, V. Model Predictive Trajectory Tracking and Collision Avoidance for Reliable Outdoor Deployment of Unmanned Aerial Vehicles. In Proceedings of the 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Madrid, Spain, 1–5 October 2018; pp. 6753–6760. [Google Scholar] [CrossRef]
  27. Ma, H.; Gao, Y.; Yang, Y.; Xu, S. Improved Nonlinear Model Predictive Control Based Fast Trajectory Tracking for a Quadrotor Unmanned Aerial Vehicle. Drones 2024, 8, 387. [Google Scholar] [CrossRef]
  28. Brescianini, D.; Hehn, M.; D’Andrea, R. Nonlinear Quadrocopter Attitude Control: Technical Report; Technical Report; ETH Zurich: Zürich, Switzerland, 2013. [Google Scholar]
  29. Werner, L.; Strohmeier, M.; Rothe, J.; Montenegro, S. Thrust vector observation for force feedback-controlled UAVs. Drones 2022, 6, 49. [Google Scholar] [CrossRef]
  30. Gonzalez-Morgado, A.; Sanchez-Cuevas, P.J.; Heredia, G.; Ollero, A. Rotor-force controller for multirotors under aerodynamic interferences. Aerosp. Sci. Technol. 2025, 157, 109861. [Google Scholar] [CrossRef]
Figure 1. Multirotors equipped with horizontal thrust-vectoring systems. The red arrows in the image show the positions of the horizontal thrusters: (a) Performing torsional work using the with 3 vectored thrusters. (b) Conducting high-pressure washing using with 3 fixed thrusters. (c) Establishing wall contact using dual thrusters (d) Enabling planar motion with a single thruster.
Figure 1. Multirotors equipped with horizontal thrust-vectoring systems. The red arrows in the image show the positions of the horizontal thrusters: (a) Performing torsional work using the with 3 vectored thrusters. (b) Conducting high-pressure washing using with 3 fixed thrusters. (c) Establishing wall contact using dual thrusters (d) Enabling planar motion with a single thruster.
Drones 09 00145 g001
Figure 2. Topologies of multirotors with thrusters: (a) Romero et al. (2007) [12], (b) Albers et al. (2010) [13], (c) Imamura et al. (2016) [15], (d) González-deSantos et al. (2020) [14], (e) Orjales et al. (2021) [19], (f) Miyazaki et al. (2022) [16], (g) Martinez et al. (2023) [17], (h) Martinez et al. (2024) [18], and (i) Martinez et al. (2024) [18].
Figure 2. Topologies of multirotors with thrusters: (a) Romero et al. (2007) [12], (b) Albers et al. (2010) [13], (c) Imamura et al. (2016) [15], (d) González-deSantos et al. (2020) [14], (e) Orjales et al. (2021) [19], (f) Miyazaki et al. (2022) [16], (g) Martinez et al. (2023) [17], (h) Martinez et al. (2024) [18], and (i) Martinez et al. (2024) [18].
Drones 09 00145 g002
Figure 3. Flight controller block diagram.
Figure 3. Flight controller block diagram.
Drones 09 00145 g003
Figure 4. Generalized topology for actuators with horizontal thruster support.
Figure 4. Generalized topology for actuators with horizontal thruster support.
Drones 09 00145 g004
Figure 5. Block diagram illustrating the communication flow within the developed system.
Figure 5. Block diagram illustrating the communication flow within the developed system.
Drones 09 00145 g005
Figure 6. Setup for torque evaluation (a). Dual-thruster configuration. (a.1) Moment evaluation with dual-thruster (b). Three-thruster configuration (b.1) Moment evaluation with 3 thrusters.
Figure 6. Setup for torque evaluation (a). Dual-thruster configuration. (a.1) Moment evaluation with dual-thruster (b). Three-thruster configuration (b.1) Moment evaluation with 3 thrusters.
Drones 09 00145 g006
Figure 7. Moment evaluation results.
Figure 7. Moment evaluation results.
Drones 09 00145 g007
Figure 8. Thrusters at a horizontal offset from the center of mass.
Figure 8. Thrusters at a horizontal offset from the center of mass.
Drones 09 00145 g008
Figure 9. Trajectory and angle setpoints during flight: (a) forward motion (+x), (b) backward motion (−x), (c) lateral right motion (+y), and (d) lateral left motion (−y).
Figure 9. Trajectory and angle setpoints during flight: (a) forward motion (+x), (b) backward motion (−x), (c) lateral right motion (+y), and (d) lateral left motion (−y).
Drones 09 00145 g009
Figure 10. Horizontal thrust-vectored thruster flight results. (a) forward motion (+x), (b) backward motion (−x), (c) lateral right motion (+y), and (d) lateral left motion (−y).
Figure 10. Horizontal thrust-vectored thruster flight results. (a) forward motion (+x), (b) backward motion (−x), (c) lateral right motion (+y), and (d) lateral left motion (−y).
Drones 09 00145 g010
Figure 11. Thrusters positioned at vertical offsets from the center of mass of the UAV.
Figure 11. Thrusters positioned at vertical offsets from the center of mass of the UAV.
Drones 09 00145 g011
Figure 12. Thrusters positioned at negative offset from the center of mass of the UAV (0 mm to −60 mm).
Figure 12. Thrusters positioned at negative offset from the center of mass of the UAV (0 mm to −60 mm).
Drones 09 00145 g012
Figure 13. Thrusters positioned at negative offset from the center of mass of the UAV (−80 mm to −100 mm).
Figure 13. Thrusters positioned at negative offset from the center of mass of the UAV (−80 mm to −100 mm).
Drones 09 00145 g013
Figure 14. Thrusters positioned at positive offset from the center of mass of the UAV (0 mm to 60 mm).
Figure 14. Thrusters positioned at positive offset from the center of mass of the UAV (0 mm to 60 mm).
Drones 09 00145 g014
Figure 15. Thrusters positioned at positive offset from the center of mass of the UAV (80 mm to 100 mm).
Figure 15. Thrusters positioned at positive offset from the center of mass of the UAV (80 mm to 100 mm).
Drones 09 00145 g015
Table 1. Topology comparison for multirotors with horizontal thrusters.
Table 1. Topology comparison for multirotors with horizontal thrusters.
SystemUAV RotorsNumber of ThrustersThrusters DirectionControl of UAV and ThrustersThruster Forces
(a)44FixedIndependent ± x B , ± y B
(b)41FixedIndependent + x B
(c)42VectorableIndependent ± x B
(d)42FixedIndependent + x B
(e)84FixedIndependent ± x B , ± y B
(f)63FixedIndependent ± x B , ± y B
(g)63VectorableIndependent ± x B , ± y B
(h)42FixedIntegrated ± x B
(i)41FixedIntegrated + x B
Table 2. Notations used in this paper.
Table 2. Notations used in this paper.
SymbolDescription
x s p R 3 Desired position vector
x ˙ s p R 3 Desired velocity vector
x ¨ s p R 3 Desired acceleration vector
ψ s p R Yaw setpoint
ψ ˙ s p R Angular velocity setpoint around the z-axis
q s p R 4 Desired attitude in quaternions
F s p R 3 Desired force vector
M s p R 3 Desired moment vector
A x y R 4 Current allocation matrix control parameter
F = [ f x , f y , f z ] T Resultant forces acting on the center of gravity
M = [ τ x , τ y , τ z ] T Resultant moments acting on the center of gravity
R R 3 × 3 Rotation Matrix
I R 3 × 3 Inertial Matrix
Θ R n Angular position of the horizontal propellers
ω i R 3 Angular velocity of the propellers
λ R 3 Vector of squared angular velocities
μ i R Lift force coefficient
κ i R Aerodynamic drag coefficient
γ i R Rotational direction where γ i = + 1 (CCW) and γ i = 1 for (CW)
Table 3. Specifications for vehicles with horizontal offsets.
Table 3. Specifications for vehicles with horizontal offsets.
Vehicle Horizontal Offset (mm)Mass (kg)Inertia (kg·m2)
2002.984 { 0.0133 , 0.0133 , 0.0249 }
3003.019 { 0.0133 , 0.0133 , 0.0249 }
4003.054 { 0.0139 , 0.0318 , 0.0430 }
5003.118 { 0.0142 , 0.0592 , 0.0701 }
Table 4. Specifications for vehicles with positive and negative vertical offsets.
Table 4. Specifications for vehicles with positive and negative vertical offsets.
Vehicle Vertical Offset (mm)Mass (kg)Inertia (kg·m2)
02.85 { 0.0134 , 0.0171 , 0.0286 }
202.85 { 0.0136 , 0.0171 , 0.0286 }
−202.85 { 0.0136 , 0.0171 , 0.0286 }
402.85 { 0.0140 , 0.0175 , 0.0286 }
−402.85 { 0.0140 , 0.0175 , 0.0286 }
602.85 { 0.0147 , 0.0182 , 0.0286 }
−602.85 { 0.0147 , 0.0182 , 0.0286 }
802.85 { 0.0157 , 0.0192 , 0.0286 }
−802.85 { 0.0157 , 0.0192 , 0.0286 }
1002.85 { 0.0169 , 0.0205 , 0.0286 }
−1002.85 { 0.0169 , 0.0205 , 0.0286 }
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Martinez, R.R.; Paul, H.; Shimonomura, K. Design and Control Strategies of Multirotors with Horizontal Thrust-Vectored Propellers. Drones 2025, 9, 145. https://doi.org/10.3390/drones9020145

AMA Style

Martinez RR, Paul H, Shimonomura K. Design and Control Strategies of Multirotors with Horizontal Thrust-Vectored Propellers. Drones. 2025; 9(2):145. https://doi.org/10.3390/drones9020145

Chicago/Turabian Style

Martinez, Ricardo Rosales, Hannibal Paul, and Kazuhiro Shimonomura. 2025. "Design and Control Strategies of Multirotors with Horizontal Thrust-Vectored Propellers" Drones 9, no. 2: 145. https://doi.org/10.3390/drones9020145

APA Style

Martinez, R. R., Paul, H., & Shimonomura, K. (2025). Design and Control Strategies of Multirotors with Horizontal Thrust-Vectored Propellers. Drones, 9(2), 145. https://doi.org/10.3390/drones9020145

Article Metrics

Back to TopTop