In the context of accelerating the construction of a maritime power, the importance of ship inspection has been increasingly emphasized. According to regulations such as the “Smart Ship Code 2023” and the “Ship Inspection Management Regulations,” underwater hull inspection is a legal requirement for ensuring the safe navigation of every vessel. This inspection is crucial for maintaining structural integrity and operational safety, as ships are constantly subjected to harsh underwater environments, including corrosion, biofouling, and physical impacts. Traditional methods, such as manual diving, pose significant safety risks, high costs, and low efficiency. Moreover, existing underwater robotic systems face challenges like poor stability, limited visibility, and inaccurate data transmission in complex aquatic conditions. To address these limitations, we have focused on the development of a bionic flapping wing robot. This bionic robot mimics the locomotion of marine organisms, such as sea turtles, to achieve enhanced stability and flexibility in underwater environments. The design leverages biomimetic principles to create a system capable of precise and efficient hull inspection, thereby reducing hazards, cutting costs, and improving detection accuracy.
Our work is driven by the need for advanced inspection technologies that can adapt to dynamic underwater settings. The bionic robot we propose is inspired by the flapping motion of sea turtle limbs, which exhibit high propulsion efficiency and maneuverability. By emulating this natural mechanism, our bionic robot aims to overcome the drawbacks of conventional underwater vehicles. In this paper, we present the comprehensive design and analysis of this bionic flapping wing robot, from its structural components to control systems and simulation experiments. We employ computational tools like FLOW Simulation to validate the design, ensuring it meets the demands of underwater hull inspection. The integration of multi-sensor fusion and path planning algorithms further enhances the robot’s autonomy and reliability. Throughout this article, we will detail each aspect of the bionic robot, emphasizing its innovative features and performance metrics.
The overall design of the bionic robot is based on biomimetic principles, specifically imitating the swimming patterns of sea turtles. Sea turtles utilize a combination of lift-based and drag-based propulsion, allowing for efficient movement in water. Our bionic robot replicates this through a flapping wing mechanism, which provides both thrust and control. The robot consists of two main parts: the body and the limb structures. The body is designed for buoyancy and maneuverability, using lightweight yet robust materials to ensure stability. The limb structures include forelimbs for propulsion and hindlimbs for balance, enabling diverse motions such as forward, backward, and turning. This dual-approach mimics biological systems, making the bionic robot highly adaptable to underwater tasks. The three-dimensional model was developed using SolidWorks, as shown in the image above, highlighting the streamlined form that reduces drag and enhances fluid dynamics.
The forelimb motion structure is central to the propulsion of the bionic robot. It features two degrees of freedom: rotation around the Z-axis for flapping and rotation around the X-axis for angle of attack adjustment. This design allows the generation of forward or backward thrust by controlling the flapping frequency and amplitude. For instance, when the fin flaps downward with a downward attack angle, it produces forward motion; conversely, an upward flap with an upward attack angle yields backward motion. Turning is achieved by opposing the attack angles on the left and right fins. A crank-link mechanism is incorporated to increase stiffness along the Z-axis, ensuring precise movement. The actuation is handled by servomotors that provide torque retention, maintaining posture against external disturbances. This forelimb design is summarized in Table 1, which outlines key parameters.
| Parameter | Value | Description |
|---|---|---|
| Degrees of Freedom | 2 | Z-axis rotation (flapping) and X-axis rotation (attack angle) |
| Flapping Frequency Range | 0.5–5 Hz | Adjustable for speed control |
| Attack Angle Range | -30° to 30° | Positive for upward, negative for downward |
| Material | Carbon Fiber | Lightweight and corrosion-resistant |
| Actuation | Servomotor | Provides precise angular control |
The hindlimb balance structure is designed for stability and adaptability. We developed two interchangeable configurations: a flat-fin hindlimb inspired by sea turtle hind flippers and a dual-tail fin hindlimb modeled after fish caudal fins. The flat-fin hindlimb aids in attitude adjustment and pitch control, offering excellent maneuverability in confined spaces. In contrast, the dual-tail fin hindlimb is optimized for high-speed swimming, with two fins oscillating 180° out of phase to cancel lateral forces and enhance stability. The fins are shaped as lunate forms to maximize propulsion efficiency. This modularity allows the bionic robot to switch between configurations based on environmental conditions, such as calm waters for precise inspection or open areas for rapid traversal. Table 2 compares the two hindlimb types.
| Configuration | Inspiration | Primary Function | Advantages | Best Use Case |
|---|---|---|---|---|
| Flat-Fin Hindlimb | Sea Turtle | Attitude adjustment and stability | High flexibility and precise control | Detailed hull inspection in complex geometries |
| Dual-Tail Fin Hindlimb | Fish Caudal Fin | High-speed propulsion | Reduced lateral forces, efficient thrust | Rapid coverage of large hull areas |
To manage buoyancy and depth control, the bionic robot incorporates a ballast system using an airbag mechanism. This system consists of a flexible airbag, tubing, a normally closed solenoid valve, an air pump, and a control board. The airbag is externally mounted on an acrylic cylinder. When the robot needs to ascend, the air pump extracts air from the cylinder into the airbag, increasing displacement and creating negative pressure in the rear chamber. For descent, the solenoid valve opens, allowing air to return to the chamber, reducing displacement. The control board regulates these actions via PWM signals from a microcontroller. Additionally, aluminum alloy weights are used to adjust the center of gravity, ensuring stable underwater navigation. The buoyancy force can be expressed by the following equation, where $\rho$ is water density, $V$ is displaced volume, and $g$ is gravitational acceleration:
$$F_b = \rho V g$$
Adjusting $V$ through the airbag system allows precise control over the bionic robot’s depth. The dynamics of the system can be modeled using Newton’s second law, incorporating drag forces and thrust from the flapping wings. The overall motion equation in the vertical direction is:
$$m \frac{d^2z}{dt^2} = F_b – mg – F_d + F_t$$
Here, $m$ is the mass of the bionic robot, $z$ is depth, $F_d$ is hydrodynamic drag, and $F_t$ is thrust from the flapping wings. By optimizing these parameters, the bionic robot achieves smooth vertical movements essential for hull scanning.
The control system of the bionic robot is a critical component that enables autonomous operation. It comprises hardware and software subsystems working in tandem. The hardware system includes a central processing unit connected to various modules: a gyroscope for orientation, the buoyancy control module, a positioning module (e.g., GPS-acoustic hybrids), and a sensor suite. The sensor suite integrates visual, laser, and inertial navigation systems (INS) to form a multi-sensor fusion platform. INS provides continuous attitude and velocity estimates but suffers from drift over time. Visual sensors, using camera modules, capture environmental point clouds and motion parameters but lack scale information. Laser sensors offer high-precision point clouds but are less intuitive. By fusing these data streams, the bionic robot compensates for individual limitations—for example, visual data corrects laser distortions, while laser data provides scale to visual inputs, and both calibrate INS drift. This fusion enhances environmental adaptability and accuracy. The hardware architecture is summarized in Table 3.
| Component | Function | Specifications |
|---|---|---|
| Microcontroller | Central processing and control | ARM Cortex-M7, 600 MHz |
| Gyroscope Module | Attitude and heading reference | 6-axis IMU (accelerometer + gyroscope) |
| Visual Sensor | Image capture and point cloud generation | 1080p camera with 120° FOV |
| Laser Scanner | High-resolution 3D mapping | 905 nm wavelength, 100 m range |
| Positioning Module | Real-time location tracking | Acoustic transponder with GPS surface link |
| Communication Interface | Data transmission and remote control | Wi-Fi and acoustic modem |
The software system is programmed in C language and includes main routines, subroutines, and wireless control programs. It handles sensor data processing, path planning, and obstacle avoidance. During operation, the bionic robot uses flapping motions to navigate while collecting hull surface data via sensors. The control system processes this data to create a 3D visual model of the hull, identifying issues like dents, corrosion, or structural defects. Algorithms for image processing combine 3D laser scanning with 2D image analysis, as shown in the workflow below. This integration allows for comprehensive detection, with results displayed as 2D charts or 3D models. The path planning employs a hybrid approach: global planning for overall inspection routes and local planning for real-time adjustments. This enables the bionic robot to avoid obstacles and adapt to currents, ensuring consistent data acquisition. The sensor fusion algorithm can be expressed using an Extended Kalman Filter (EKF), where the state vector $\mathbf{x}$ includes position, velocity, and attitude:
$$\mathbf{x}_{k} = f(\mathbf{x}_{k-1}, \mathbf{u}_{k-1}) + \mathbf{w}_{k-1}$$
$$\mathbf{z}_{k} = h(\mathbf{x}_{k}) + \mathbf{v}_{k}$$
Here, $f$ is the state transition function, $h$ is the measurement function, $\mathbf{w}$ and $\mathbf{v}$ are process and measurement noise, respectively. By updating $\mathbf{x}$ with sensor inputs, the bionic robot maintains accurate localization. Additionally, the control law for flapping wing actuation is derived from PID (Proportional-Integral-Derivative) principles, with the output torque $\tau$ given by:
$$\tau = K_p e(t) + K_i \int_0^t e(\tau) d\tau + K_d \frac{de(t)}{dt}$$
where $e(t)$ is the error between desired and actual wing positions, and $K_p$, $K_i$, $K_d$ are tuned gains. This ensures precise motion control for the bionic robot during inspection tasks.
To validate the design, we conducted simulation experiments using FLOW Simulation software. The first experiment analyzed the effect of flow velocity on surface pressure distribution during flapping motion. We modeled a fluid environment with boundary conditions and set flow velocities ranging from 0 to 2 knots. A custom program was written to simulate the kinematic model of the flapping wing, with periodic motion parameters. The results, as shown in Table 4, indicate that as flow velocity increases, the flapping amplitude decreases inversely. High-pressure zones on the wing surface are concentrated near the edges at peak flapping points, suggesting these areas generate significant reactive forces for propulsion. The pressure distribution $P$ over the wing surface can be described by the Bernoulli equation, modified for unsteady flow:
$$P + \frac{1}{2} \rho v^2 + \rho g h = \text{constant}$$
where $v$ is local fluid velocity, and $h$ is depth. The simulation confirms that the bionic robot’s wings efficiently convert fluid dynamics into thrust, crucial for underwater mobility.
| Flow Velocity (knots) | Flapping Amplitude (degrees) | Max Surface Pressure (Pa) | High-Pressure Zone Location |
|---|---|---|---|
| 0.5 | 45 | 1250 | Wing edge, mid-flap |
| 1.0 | 30 | 2100 | Wing edge, peak flap |
| 1.5 | 20 | 2900 | Wing edge, peak flap |
| 2.0 | 15 | 3500 | Wing edge, peak flap |
The second experiment assessed the stability of the bionic robot during hull inspection, comparing it with a mainstream underwater robot. We simulated a scenario with a 3,000-ton bulk carrier docked in water, introducing random environmental disturbances like currents and turbidity. Both robots were tasked with scanning the hull using 3D laser scanning and 2D image fusion. The bionic robot, with its flapping wing design, demonstrated superior stability due to its ability to quickly adjust posture. In contrast, the conventional robot suffered from reduced flexibility, leading to collisions and data inaccuracies. The point cloud data accuracy was quantified using the root mean square error (RMSE) between measured and actual hull surfaces, defined as:
$$\text{RMSE} = \sqrt{\frac{1}{n} \sum_{i=1}^n (y_i – \hat{y}_i)^2}$$
where $y_i$ are actual points, $\hat{y}_i$ are measured points, and $n$ is the number of samples. The bionic robot achieved an RMSE of 2.1 mm, while the mainstream robot had 5.8 mm, highlighting the enhanced precision of our design. This stability stems from the damped dynamics of the flapping system, which can be modeled as a second-order system:
$$\frac{d^2\theta}{dt^2} + 2\zeta\omega_n \frac{d\theta}{dt} + \omega_n^2 \theta = u(t)$$
Here, $\theta$ is wing angle, $\zeta$ is damping ratio, $\omega_n$ is natural frequency, and $u(t)$ is control input. The bionic robot’s parameters are tuned for critical damping ($\zeta \approx 1$), minimizing oscillations and ensuring smooth motion. These experiments verify that the bionic robot meets the demands of underwater hull inspection, offering robust performance in challenging conditions.
In conclusion, we have presented the design and analysis of a bionic flapping wing robot for underwater hull inspection. This bionic robot leverages biomimetic principles, imitating sea turtle locomotion to achieve high stability and flexibility. The structural design includes forelimbs for propulsion and interchangeable hindlimbs for balance, supported by a ballast system for depth control. The control system integrates multi-sensor fusion and advanced algorithms for autonomous operation. Simulation experiments using FLOW Simulation confirm that the bionic robot effectively manages flow interactions and maintains superior stability compared to conventional systems. The results demonstrate that this bionic robot can perform accurate hull inspections, reducing risks and costs while improving efficiency. Future work will involve prototyping and real-world testing to further refine the bionic robot’s capabilities, potentially expanding its applications to other underwater tasks such as pipeline monitoring or environmental surveys. Throughout this project, the emphasis on biomimicry has proven essential for developing a robust and adaptable bionic robot, underscoring the value of nature-inspired engineering in advancing maritime technology.