In recent years, the field of bionic robotics has seen tremendous growth, driven by the desire to emulate the efficiency and adaptability of biological organisms. As a researcher in this domain, I have been particularly fascinated by underwater bionic robots, which offer unique advantages over traditional rigid counterparts. These bionic robots, inspired by marine life, promise enhanced maneuverability, energy efficiency, and minimal environmental disturbance. Among various inspirations, the jellyfish stands out due to its simple yet effective propulsion mechanism, making it an ideal model for developing advanced underwater bionic robots. In this work, we detail the design, fabrication, and control of a bionic robot that mimics the Aurelia jellyfish, utilizing shape memory alloy (SMA) actuator modules. Our goal is to create a bionic robot capable of multi-pattern swimming in three-dimensional space, with applications in marine exploration, environmental monitoring, and search-and-rescue operations. This bionic robot represents a significant step forward in soft robotics, showcasing how bio-inspired designs can lead to innovative solutions.
The core of our bionic robot lies in the SMA actuator modules, which function as artificial muscles. SMA wires, known for their ability to recover predefined shapes upon heating, offer high force-to-weight ratios and compact form factors, making them suitable for bionic robot applications. We designed these modules to replicate the contraction and relaxation of jellyfish bell muscles. Each module consists of a driving layer with SMA wires, a restoring layer of spring steel, and a filling layer of polydimethylsiloxane (PDMS). The fabrication process involves threading SMA wires onto a PCB, molding with PDMS, and attaching the spring steel. To optimize performance, we applied a pre-tension of 0.5% to the SMA wires during assembly, ensuring maximal bending angles. The modules measure 100 mm in length, 10 mm in width, and 2.5 mm in thickness, with an effective SMA length of 87 mm. The electrical characteristics were tested extensively, revealing an average resistance of 6 Ω with less than 0.5 Ω variation. We conducted experiments to determine the optimal driving voltage by analyzing bending angles and response times under different voltages. The results, summarized in Table 1, indicate that a voltage of 20 V yields the best performance, with a bending angle of 150° and an actuation frequency of up to 2.48 Hz. This performance is critical for the bionic robot’s swimming capabilities, as it directly influences propulsion efficiency.
| Driving Voltage (V) | Max Bending Angle (°) | Time to Max Angle (s) | Cooling Time to Initial State (s) | Observed Phenomena |
|---|---|---|---|---|
| 15 | 120 | 0.8 | 1.5 | Slow response, no overshoot |
| 20 | 150 | 0.5 | 1.2 | Optimal balance, minimal overshoot |
| 25 | 160 | 0.3 | 2.0 | Overshoot and vibration, residual stress |
| 30 | 165 | 0.2 | 2.5 | Severe overshoot, thermal accumulation |
The bending dynamics of the SMA actuator module can be modeled using thermodynamic principles. The strain recovery of SMA wires upon heating is governed by the phase transformation from martensite to austenite. The relationship between temperature, stress, and strain can be expressed as:
$$ \sigma = E(\epsilon – \epsilon_L \xi) + \Theta (T – T_0) $$
where \(\sigma\) is the stress, \(E\) is the Young’s modulus, \(\epsilon\) is the strain, \(\epsilon_L\) is the maximum recoverable strain, \(\xi\) is the martensite fraction, \(\Theta\) is the thermal expansion coefficient, \(T\) is the temperature, and \(T_0\) is the reference temperature. The martensite fraction \(\xi\) evolves with temperature according to:
$$ \xi = \frac{1}{2} \cos\left[a_M (T – M_f)\right] + \frac{1}{2} \quad \text{for cooling} $$
$$ \xi = \frac{1}{2} \cos\left[a_A (T – A_s)\right] + \frac{1}{2} \quad \text{for heating} $$
Here, \(a_M\) and \(a_A\) are material constants, \(M_f\) is the martensite finish temperature, and \(A_s\) is the austenite start temperature. For our bionic robot, we used SMA wires with a phase transition temperature of 90°C and a maximum strain recovery of 4.5%. The force generated by the SMA module during bending contributes to the propulsion of the bionic robot. The torque \(\tau\) produced by the module can be approximated as:
$$ \tau = F \cdot d = (\sigma A) \cdot d $$
where \(F\) is the force, \(A\) is the cross-sectional area of the SMA wire, and \(d\) is the distance from the SMA wire to the neutral axis of the module. In our design, \(d = 1.5\) mm, and with two SMA wires of diameter 0.15 mm each, the total cross-sectional area is \(A = 2 \times \pi (0.075)^2 \approx 0.0353 \, \text{mm}^2\). This torque drives the bending motion, which is essential for the jet propulsion mechanism in our bionic robot.
Building on these actuator modules, we constructed the bionic robot by assembling six modules in a radially symmetric configuration, mimicking the jellyfish’s bell structure. The robot features a central housing that contains the power supply and control circuitry, surrounded by a flexible silicone skin that forms the bell. The overall design ensures buoyancy and stability, with a weight of 512 g and a density of 1.031 g/cm³, close to that of water. In its relaxed state, the bionic robot has a diameter of 210 mm and a height of 80 mm, while during contraction, these dimensions reduce to 142 mm and 88 mm, respectively. This compact and lightweight structure enables efficient swimming, making our bionic robot suitable for underwater exploration. The use of modular SMA actuators allows for scalability and ease of maintenance, which are key advantages for deploying bionic robots in real-world environments.
To achieve coordinated swimming motions, we implemented a control system based on central pattern generators (CPGs). CPGs are neural networks that produce rhythmic outputs without sensory feedback, making them ideal for controlling the periodic contractions of a bionic robot. We adopted a nonlinear oscillator model, inspired by biological CPGs, to generate control signals for the six SMA actuator modules. The state equations for each oscillator \(i\) are defined as:
$$ \dot{\theta}_i = 2\pi f_i + \sum_{j} r_j w_{ij} \sin(\theta_j – \theta_i – \phi_{ij}) $$
$$ \ddot{r}_i = a_i \left[ \frac{a_i}{4} (R_i – r_i) – \dot{r}_i \right] $$
$$ x_i = r_i \cos(\theta_i) $$
$$ S_i = \begin{cases} 0 & \text{if } x_i \leq 0 \\ 1 & \text{if } x_i > 0 \end{cases} $$
where \(\theta_i\) and \(r_i\) are the phase and amplitude state variables, \(f_i\) and \(R_i\) are the desired frequency and amplitude, \(w_{ij}\) and \(\phi_{ij}\) are the coupling weights and phase biases between oscillators, \(a_i\) is a positive constant determining the convergence rate, and \(S_i\) is the binary output used to actuate the SMA modules. For our bionic robot, we set \(w_{ij} = 200\) and \(a_i = 30\) for stability, while \(f_i\) and \(\phi_{ij}\) are tunable parameters to achieve different swimming patterns. The CPG network topology, shown in Figure 1, connects six oscillators in a ring configuration, enabling synchronized or phase-shifted actuation. This model allows our bionic robot to perform various maneuvers, such as vertical swimming and turning, by simply adjusting the frequency and phase parameters. The effectiveness of this control approach is demonstrated through experiments, confirming that CPG-based control is a robust method for bionic robot locomotion.
We conducted extensive experiments to evaluate the swimming performance of our bionic robot. The tests were performed in a water tank, and motion was captured using high-speed cameras for analysis. We focused on two primary swimming modes: vertical ascent and turning. For vertical swimming, we varied the actuation frequency from 0.5 Hz to 2.5 Hz and measured the average swimming speed and duration. The results, presented in Table 2, show that the bionic robot achieves a maximum speed of 5.28 cm/s at 2 Hz, but with reduced operational time due to thermal accumulation in the SMA modules. At 1 Hz, the robot maintains a steady speed of 2.15 cm/s for over 120 seconds, indicating a balance between performance and sustainability. These findings highlight the trade-offs in designing bionic robots with SMA actuators, where higher frequencies boost speed but may lead to overheating. The swimming motion follows a jet propulsion cycle: contraction of the bell expels water downward, generating thrust, and relaxation allows water re-entry, propelled by the stored elastic energy in the silicone skin and spring steel. This cycle is analogous to that of natural jellyfish, underscoring the bio-inspired efficiency of our bionic robot.
| Frequency (Hz) | Average Speed (cm/s) | Swimming Duration (s) | Observations |
|---|---|---|---|
| 0.5 | 1.05 | Stable, low power consumption | |
| 1.0 | 2.15 | Optimal for sustained operation | |
| 1.5 | 3.78 | 60 | Moderate thermal effects |
| 2.0 | 5.28 | 30 | High speed, rapid thermal buildup |
| 2.5 | 5.50 | 15 | Severe overheating, performance degradation |
For turning maneuvers, we exploited differential actuation by varying frequencies between opposite sides of the bionic robot. When three SMA modules on one side were kept static while the other three operated at 2 Hz, the robot achieved a minimum turning radius of 31.95 cm. This demonstrates the agility of our bionic robot, enabling it to navigate complex underwater environments. The turning dynamics can be analyzed using fluid mechanics principles. The thrust force \(F_t\) generated during contraction is given by:
$$ F_t = \rho A_j v_j^2 $$
where \(\rho\) is the water density, \(A_j\) is the cross-sectional area of the jet exit, and \(v_j\) is the jet velocity. For asymmetric actuation, the net torque \(T_{net}\) causing turning is:
$$ T_{net} = F_t \cdot d_{offset} $$
with \(d_{offset}\) being the offset distance from the center of mass. This torque leads to angular acceleration \(\alpha\) according to:
$$ \alpha = \frac{T_{net}}{I} $$
where \(I\) is the moment of inertia of the bionic robot. By controlling the phase differences in the CPG network, we can modulate \(F_t\) and \(d_{offset}\) to achieve desired turning behaviors. Table 3 summarizes the turning performance under different actuation patterns, illustrating how CPG parameters influence maneuverability. These results validate the versatility of our bionic robot, showcasing its potential for adaptive swimming in diverse scenarios.
| Actuation Pattern | Phase Bias (rad) | Turning Radius (cm) | Angular Velocity (rad/s) |
|---|---|---|---|
| Symmetric (all modules at 1 Hz) | 0 | ∞ (straight line) | 0 |
| Asymmetric (3 modules static, 3 at 1 Hz) | π | 50.2 | 0.42 |
| Asymmetric (3 modules static, 3 at 2 Hz) | π | 31.95 | 0.85 |
| Gradual phase shift (all at 1 Hz) | π/6 | 75.3 | 0.21 |
The development of this bionic robot also involved addressing challenges such as thermal management and energy efficiency. SMA actuators are prone to heat accumulation, which can lead to performance degradation or failure—a phenomenon known as thermal deadlock. To mitigate this, we incorporated passive cooling through the surrounding water and optimized the actuation cycles. The power consumption \(P\) of each SMA module can be calculated as:
$$ P = \frac{V^2}{R} $$
where \(V\) is the driving voltage and \(R\) is the resistance. With \(V = 20\) V and \(R = 6\) Ω, each module consumes approximately 66.67 W during activation. For six modules, the total peak power is 400 W, but due to intermittent actuation, the average power is lower, around 100 W. This energy demand is supplied by a battery pack of six IMR18650 cells, providing a nominal voltage of 20.6 V and a capacity of 3000 mAh. The bionic robot can operate for about 30 minutes on a single charge, depending on the swimming mode. Future improvements could involve integrating thermal sensors and adaptive control algorithms to dynamically adjust actuation patterns, enhancing the longevity of the bionic robot. Such advancements are crucial for deploying bionic robots in long-duration missions, where reliability is paramount.
In comparison to other underwater bionic robots, our design offers several advantages. Many existing bionic robots, such as those inspired by fish or octopuses, rely on complex mechanisms or external power sources, limiting their autonomy and scalability. Our bionic robot, with its modular SMA actuators and CPG-based control, achieves a high degree of bio-mimicry while maintaining simplicity. For instance, the use of silicone skin for the bell provides flexibility and durability, reducing the risk of damage in rugged environments. Additionally, the radial symmetry of the actuator modules ensures balanced thrust, minimizing rotational instability during swimming. These features make our bionic robot a promising platform for research and applications in marine robotics. The success of this bionic robot also underscores the potential of SMA-based actuation in soft robotics, where traditional motors and gears are often unsuitable due to size or weight constraints.
Looking ahead, there are numerous opportunities to enhance this bionic robot. One direction is to incorporate sensory feedback for closed-loop control, allowing the bionic robot to respond to environmental stimuli such as obstacles or currents. This could be achieved by integrating pressure sensors, accelerometers, or cameras, coupled with machine learning algorithms for real-time decision-making. Another area is material innovation; for example, using SMA wires with lower phase transition temperatures could reduce power consumption and thermal issues. Furthermore, scaling the design to larger or smaller sizes could expand its applicability, from micro-robots for medical procedures to macro-robots for oceanography. We also envision swarm robotics, where multiple bionic robots collaborate to perform complex tasks, leveraging CPG networks for coordinated movements. These advancements will push the boundaries of what bionic robots can achieve, paving the way for more intelligent and autonomous systems.
In conclusion, this work presents a comprehensive approach to developing a bionic robot inspired by the jellyfish. Through the design of SMA actuator modules, the implementation of a CPG control system, and extensive experimental validation, we have demonstrated a functional and versatile underwater bionic robot. The robot achieves multi-pattern swimming, including vertical ascent and turning, with performance metrics that rival those of natural jellyfish. The use of bio-inspired principles not only enhances efficiency but also promotes sustainability, as seen in the energy-saving actuation cycles. As bionic robots continue to evolve, they hold immense promise for transforming underwater exploration, offering solutions that are both innovative and ecologically friendly. Our bionic robot serves as a testament to the power of biomimicry in robotics, highlighting how nature’s designs can inspire cutting-edge technology. We believe that this bionic robot will contribute to the growing field of soft robotics, encouraging further research into adaptive and resilient machines.