The application of unmanned aerial vehicle (UAV) technology has seen remarkable advancements in intelligence in recent years, finding widespread use in military, commercial, and scientific research fields. The quadrotor drone, with its simple structure and vertical take-off and landing capabilities, is a particularly versatile platform. However, when tasked with transporting external payloads, the quadrotor drone faces significant challenges. The added mass increases the system’s inertia, leading to sluggish attitude response and reduced resistance to disturbances. This degradation in flight performance limits the drone’s agility and precision, especially in applications requiring delicate manipulation or operation in dynamic environments.
To address these inherent problems, researchers globally have focused on designing intelligent cable-suspended load systems. Traditional methods for payload handling primarily involve rigid grippers or rigidly attached cargo bays. While functional, these methods directly couple the payload’s inertia to the airframe, exacerbating the control challenges mentioned. An alternative approach, cable-suspension, offers a promising solution. By suspending the payload via a flexible cable, the direct inertial coupling is decoupled to some extent, potentially preserving more of the quadrotor drone‘s innate agility. Nevertheless, this method introduces its own complexities, such as managing the oscillatory dynamics of the pendulum-like load and ensuring stable control amidst external disturbances.
This article presents the design of an intelligent cable-suspended payload system for a quadrotor drone, investigating the stability of the combined drone-payload system. The core innovation lies in a hierarchical hardware architecture and integrated software control strategy designed to mitigate the negative effects of the suspended load. The hardware system is built around a perception controller for high-level navigation and a dedicated flight controller for low-level stabilization. The software integrates obstacle avoidance, trajectory tracking, and a specialized suspension control algorithm. This design enables the precise and stable control of the payload’s position and attitude at the end of the cable, allowing for the deployment of end-effectors for remote operations. The following sections detail the overall system architecture, hardware composition, and software design of this cable-suspended system.
1. Overall System Architecture
The proposed system consists of a quadrotor drone carrying a payload suspended by a flexible cable. This configuration, illustrated conceptually, aims to reduce the direct impact of the payload’s inertia on the drone’s core flight dynamics. The primary objective is to control the attitude and position of the terminal payload precisely, enabling it to perform specialized tasks.
The system employs a hierarchical design, as shown in the block diagram below, separating high-level perception and planning from low-level motion control. This separation allows for complex environmental processing without overburdening the time-critical flight control loops.
| Layer | Core Component | Primary Function | Key Sensors/Actuators |
|---|---|---|---|
| Upper Layer (Perception & Navigation) | Jetson Nano | Environmental perception, target recognition, global/local path planning, suspension control algorithm. | Depth Camera, 2D LiDAR. |
| Lower Layer (Flight Control & Stabilization) | STM32F103 Microcontroller | Attitude stabilization, height control, motor mixing, real-time sensor fusion. | IMU (Gyro/Accel), Barometer, Ultrasonic Sensor, Optical Flow, PWM outputs to ESCs. |
| Communication | UART Serial Link | Bidirectional data exchange: sending navigation commands down and receiving state feedback up. | – |
| Payload | Terminal Module | Hosts end-effector; contains its own IMU and positioning device for state feedback. | Payload IMU, Payload Positioning Marker/Sensor. |
The upper-layer perception controller, centered on a Jetson Nano, processes data from a depth camera and a 2D LiDAR. It handles tasks such as obstacle detection, target recognition, and local path re-planning for collision avoidance. It also runs the high-level suspension control logic. The lower-layer flight controller, based on an STM32 microcontroller, is responsible for the fundamental stability of the quadrotor drone. It fuses data from an Inertial Measurement Unit (IMU), barometer, ultrasonic sensor, and optical flow sensor to estimate the drone’s state (attitude, height, velocity). A high-speed UART serial link facilitates communication between the two layers, allowing the Jetson Nano to send attitude and throttle commands while receiving real-time state feedback from the flight controller.
2. Hardware System Composition
The hardware system is divided into two synergistic subsystems: the Navigation & Perception System and the Flight Control & Stabilization System.
2.1 Navigation & Perception System Design
This system is responsible for enabling autonomous navigation and interaction with the environment. Its functional pipeline can be described as follows:
- Obstacle Avoidance Path Planner: Takes the global waypoints (mission plan) and real-time obstacle coordinates from the perception sensors (camera/LiDAR) as input. It generates a safe, feasible local trajectory that avoids collisions.
- Trajectory Tracking Controller: Uses the drone’s current position (from GPS or visual odometry) as feedback. It takes the local trajectory from the planner and outputs desired attitude and thrust commands to guide the drone precisely along the path.
- Height Controller: Maintains a stable altitude. It uses a compensated height estimate (corrected for attitude-induced errors) as feedback and a desired height setpoint as input to compute vertical thrust adjustments.
The final output of this perception-navigation pipeline is a set of desired attitude angles and a collective thrust command, which are sent to the lower-level flight controller.
2.2 Flight Control System Design
This is the core stabilization system for the quadrotor drone. Its hardware foundation and control logic are critical for managing the added complexity of a suspended load.
| Component Category | Specific Sensor/Module | Measurement/Function |
|---|---|---|
| State Estimation Sensors | 6-Axis IMU (Gyroscope + Accelerometer) | Angular velocity and linear acceleration in body frame. |
| Barometer | Atmospheric pressure for altitude estimation. | |
| Ultrasonic Sensor / Optical Flow | Ground-relative height and horizontal velocity (especially in GPS-denied areas). | |
| Processing Core | STM32F103 Microcontroller | Runs sensor fusion algorithms (e.g., Complementary Filter), executes PID control loops, performs motor mixing. |
| Actuation Interface | Electronic Speed Controllers (ESCs) | Converts PWM signals from the flight controller into precise motor speeds. |
The flight control software is structured around cascaded PID (Proportional-Integral-Derivative) control loops. Two primary control loops are implemented:
- Attitude Stabilization Loop (Inner Loop): This is a fast, high-frequency loop that directly stabilizes the drone’s angular rates. It uses feedback from the gyroscope to ensure the drone responds quickly and dampens disturbances. The control law for, say, the roll axis rate can be expressed as:
$$ \Omega_{cmd} = K_{p,\omega}(\omega_{des} – \omega_{meas}) + K_{i,\omega}\int(\omega_{des} – \omega_{meas})dt + K_{d,\omega}\frac{d}{dt}(\omega_{des} – \omega_{meas}) $$
where $\Omega_{cmd}$ is the corrective motor command for roll, $\omega_{des}$ is the desired angular rate (from the outer loop), and $\omega_{meas}$ is the measured rate. - Attitude & Position Loop (Outer Loop): This slower loop controls the drone’s orientation (Euler angles: roll $\phi$, pitch $\theta$, yaw $\psi$) and height. It takes desired angles/height from the navigation layer or remote controller and uses feedback from the attitude estimator (which fuses IMU and other data) to compute desired angular rates for the inner loop. The angle PID controller for roll is:
$$ \omega_{des,\phi} = K_{p,\phi}(\phi_{des} – \phi_{est}) + K_{i,\phi}\int(\phi_{des} – \phi_{est})dt + K_{d,\phi}\frac{d}{dt}(\phi_{des} – \phi_{est}) $$
Similar structures exist for pitch, yaw, and height (which outputs a collective thrust adjustment).
Finally, a “mixer” translates the four control outputs (thrust $T$, roll torque $\tau_{\phi}$, pitch torque $\tau_{\theta}$, yaw torque $\tau_{\psi}$) into individual motor speeds for the quadrotor drone‘s X-configuration:
$$
\begin{bmatrix}
\omega_1^2 \\
\omega_2^2 \\
\omega_3^2 \\
\omega_4^2
\end{bmatrix}
=
\begin{bmatrix}
1 & -1 & -1 & 1 \\
1 & -1 & 1 & -1 \\
1 & 1 & -1 & -1 \\
1 & 1 & 1 & 1
\end{bmatrix}
\begin{bmatrix}
T \\
\tau_{\phi} \\
\tau_{\theta} \\
\tau_{\psi}
\end{bmatrix}
$$
where $\omega_i$ is the rotational speed of motor $i$, and the matrix coefficients depend on thrust and torque constants.
3. Software System and Control Algorithm Design
The software integrates the functionalities of both hierarchical layers to achieve stable cable-suspended flight. The core challenge is designing a controller that not only stabilizes the quadrotor drone but also manages the pendulum dynamics of the payload.
3.1 Cable-Suspended Payload Control System
The complete system, comprising the drone, cable, and payload, is an underactuated system (the drone has 4 control inputs, but the payload adds more degrees of freedom). Our control strategy is a cascaded design, as shown in the control flow diagram.
The payload is equipped with its own IMU and a visual marker (or other positioning device like ultra-wideband), allowing the upper-layer perception controller to estimate the payload’s attitude $(\phi_p, \theta_p)$ and position $(x_p, y_p, z_p)$ relative to the drone or a global frame.
- Payload Position Controller: This outer-loop controller aims to make the payload track a desired trajectory $(x_{p,des}, y_{p,des}, z_{p,des})$. Using the payload’s position error, it calculates a required drone position or velocity. Since the drone’s horizontal position is controlled by its attitude, this ultimately translates into desired roll and pitch angles for the drone. A simplified horizontal control law could be:
$$ \phi_{des} = K_{p,x}(x_{p,des} – x_p) + K_{d,x}(\dot{x}_{p,des} – \dot{x}_p) $$
$$ \theta_{des} = K_{p,y}(y_{p,des} – y_p) + K_{d,y}(\dot{y}_{p,des} – \dot{y}_p) $$
The sign depends on the coordinate system convention. - Payload Attitude Controller: This controller manages the orientation of the payload itself (e.g., keeping a camera level or orienting a gripper). It uses the payload’s IMU feedback to calculate a corrective torque. However, since we cannot directly torque the payload, this desired torque is achieved by inducing specific accelerations in the quadrotor drone. For small angles, a desired payload attitude $\phi_{p,des}$ can be achieved by commanding a drone attitude that lags or leads appropriately, effectively using the cable’s tension to apply a corrective moment to the payload.
The outputs from these two payload-centric controllers—desired drone attitude $(\phi_{des}, \theta_{des}, \psi_{des})$ and collective thrust $T_{des}$—are then passed as setpoints to the drone’s own inner-loop attitude and height controllers described in Section 2.2. This cascaded structure allows for coherent control of the combined system.
3.2 Integrated Autopilot and System Dynamics
The full flight control system integrates the standard quadrotor drone autopilot with the suspension control commands. The control architecture is a multi-loop cascade ensuring stability from the fast motor dynamics up to the slower payload position dynamics.
The dynamics of the quadrotor drone can be simplified for controller design. Let $m_q$ be the drone mass and $I$ its inertia matrix. The translational dynamics under small angles are:
$$ \ddot{x} = \frac{T}{m_q} (\sin\psi \sin\phi + \cos\psi \sin\theta\cos\phi) $$
$$ \ddot{y} = \frac{T}{m_q} (-\cos\psi \sin\phi + \sin\psi \sin\theta\cos\phi) $$
$$ \ddot{z} = \frac{T}{m_q} (\cos\theta \cos\phi) – g $$
where $g$ is gravity. The rotational dynamics are governed by:
$$ I \dot{\boldsymbol{\omega}} + \boldsymbol{\omega} \times (I \boldsymbol{\omega}) = \boldsymbol{\tau} $$
where $\boldsymbol{\omega}$ is the angular velocity vector and $\boldsymbol{\tau}$ is the torque vector.
The suspended payload, modeled as a point mass $m_p$ at cable length $l$, introduces coupled dynamics. The swing angles $(\alpha, \beta)$ define the payload’s deviation. The forces on the payload are tension and gravity. The coupled drone-payload equations are complex, but for controller synthesis, a linearized model about the hover state (drone level, payload hanging vertically) is often used. The interaction shows that drone accelerations $\ddot{x}_q, \ddot{y}_q$ directly excite the payload swing dynamics:
$$ l \ddot{\alpha} + \ddot{x}_q \cos\alpha + g \sin\alpha \approx 0 $$
$$ l \ddot{\beta} + \ddot{y}_q \cos\beta + g \sin\beta \approx 0 $$
(for small angles, $\cos\alpha \approx 1$, $\sin\alpha \approx \alpha$).
The integrated autopilot must therefore be designed with damping to suppress these induced oscillations. This is achieved by incorporating feedback from the payload state (swing angles or rates) into the drone’s attitude or rate controllers, effectively adding a damping term to the swing dynamics. For instance, a term proportional to the payload’s swing rate $\dot{\alpha}$ can be added to the drone’s desired pitch rate command:
$$ \omega_{des,\theta} = PID(\theta_{error}) – K_{damp} \cdot \dot{\alpha} $$
This active damping is crucial for stabilizing the quadrotor drone cable-suspended system during aggressive maneuvers or in windy conditions.
4. Conclusion and Future Work
Traditional methods of rigidly attaching payloads to a quadrotor drone significantly impair its maneuverability and flight performance due to increased inertia and altered dynamics. The cable-suspension approach offers a compelling alternative for transporting payloads, particularly when terminal positioning and orientation of the load are required. This article presented the comprehensive design of such a system, featuring a hierarchical hardware architecture with a dedicated perception controller and a robust flight controller. The software framework integrates cascaded PID control for drone stabilization with specialized algorithms for payload position and attitude regulation.
The key contribution is the formulation of a control strategy that enables a quadrotor drone to precisely and stably control both its own attitude and the state of a payload suspended via a flexible cable. By addressing the swing dynamics through state feedback and active damping, the system mitigates the problems of inertia-induced sluggishness and disturbance susceptibility. This design lays a practical foundation for deploying quadrotor drones in applications requiring remote manipulation, such as sensor deployment, sample collection, or light assembly tasks in hard-to-reach locations. Future work will focus on experimental validation of the control algorithms, robustness testing in windy conditions, and extending the system to handle varying cable lengths and payload masses autonomously.