Next Article in Journal
A Transit Tilt and Offset Errors Calibration Method for Improving Laser Tracker Measurement Accuracy Based on the Telecentric Measurement System
Previous Article in Journal
Dual-Targeting Gold Nanoparticles: Simultaneous Decoration with Ligands for Co-Transporters SGLT-1 and B0AT1
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Octopus Predation-Inspired Underwater Robot Capable of Adsorption through Opening and Closing Claws

College of Mechanical and Electrical Engineering, Harbin Engineering University, No. 145 Nantong Street, Nangang District, Harbin 150001, China
*
Author to whom correspondence should be addressed.
Appl. Sci. 2024, 14(6), 2250; https://doi.org/10.3390/app14062250
Submission received: 30 January 2024 / Revised: 26 February 2024 / Accepted: 5 March 2024 / Published: 7 March 2024

Abstract

:
Underwater unmanned robots are an essential tool for human underwater exploration and detection and are widely employed in a variety of underwater operational settings. One of the hottest issues in this field is applying bionic notions to the creation of underwater unmanned robots by simulating fish swimming or cephalopod crawling. Using the tentacle suction cup adsorption technique during octopus’ predation as a model, underwater magnetic adsorption robots with the opening and closing claws were studied in this paper. First, the robot’s general structural design is presented. The claw mechanism is demonstrated by mimicking the octopus’s tentacle action during feeding, which primarily consists of an opening and closing claw that replicates the octopus’s tentacle and a magnetic adsorption unit that replicates the octopus’s suction cup adsorption. Then, the Kriging response surface optimization method is used to optimize the design of the claw mechanism to obtain excellent mechanical properties, and simulation software is used to verify. Finally, a robot prototype was built and its pool tests were conducted, with some experimental results presented. The experimental results show that after the robot reaches the predetermined position through pneumatic ejection and secondary propulsion launch, it can quickly open its claws within 0.11 s and apply 462.42 N adsorption force to complete the adsorption of the target.

1. Introduction

The oceans are a valuable resource for the long-term sustainability of human society, and many nations are investing more money in ocean mapping and the production of naval hardware [1]. In addition to being a vital instrument for conducting underwater operations and exploring unexplored oceanic regions, underwater unmanned robots are also essential for researching marine biology, seabed topography mapping, and marine resource discovery [2,3,4,5]. As bionic theory has advanced, an increasing number of underwater robots built on bionic principles have been created; bionic mechanical fish, for instance, with many propulsion modes, have been created by mimicking fish moving in water [6,7,8]. Additionally, several soft underwater robots with good deformation properties have been built based on bionic notions to enable underwater robots to adapt to complicated underwater settings and accomplish some demanding tasks [9,10].
When it comes to bionic underwater robot design, octopuses have been preferred. Numerous high-performing underwater robots have been influenced by octopuses. F. Giorgio-Serchi et al. developed an underwater soft robot with underwater legged mobility and aquatic pulse-jet propulsion, modeled after an octopus. The robot is made up of a platform that can move in several directions by attaching swimming and crawling elements to it. A pulse jet thruster is used to propel the device through the water, and four flexible limbs are used for leg movement [11]. Similar soft robots, which emulate the tentacle crawling of octopuses, were employed by Federico Renda et al. in their kinetic modeling of underwater soft robotics [12]. An eight-armed soft robot was created by Cianchetti et al., and its performance was assessed by experiments. The bionic octopus idea was applied by an eight-armed soft robot, which combines eight radially extending limbs with a central body housing primary processing units. The primary functions of the forearms are grabbing and extending, whereas the other muscles are involved in movement. The outcomes of the experiment demonstrate that the robot modeled like an octopus can walk in water by employing the same technique as the animal model and that, because of its conical form and soft arm material, it can grip things of all sizes and shapes [13]. Inspired by the gripping behavior of octopuses, Calisti and Giorelli, et al. presented a solution for grasping and movement in soft robots. The octopus arm muscles were mimicked by fabricating a silicone arm with wires inserted in it. The new arm can move in a propulsion-based manner and grip items, like the crawling motions of an octopus [14]. At present, the majority of octopus-inspired underwater robot designs mimic the movement of their tentacles to enable them to adapt to the complex underwater environment. Only a small percentage of these designs refer to the tentacles’ flexibility to design robotic hands that perform well in gripping, and very few mimic the octopus suction cup adsorption of biomimetic design.
This paper presents the design of an octopus predation-inspired underwater robot capable of adsorption through opening and closing claws. The design was inspired by the opening tentacles of octopuses, which employ suction cups to adsorb food as they feed on prey. The robot’s claw mechanism is made up of two parts: the design of the opening and closing claws, which mimics the octopus’s tentacles, and the design of the magnetic adsorption unit, which is modeled after the suction cups found on the octopus’s tentacles. The basic construction of the robot, which has an overall length of around 370 mm, is made up of three components: the propulsion module, the power module, and the claw mechanism. The interior of the capsule can also hold mission payload systems, such as reconnaissance modules. An underwater unmanned vehicle (UUV) moves the robot to a prearranged location, and the UUV’s launcher propels it toward the target. The robot then extends its claws to connect to the target after being driven a certain distance by its power unit, the propulsion module. The functional modules installed on the robot can accomplish activities like reconnaissance and detection during the synchronized movement of the robot toward the destination, following the completion of adsorption. The robots’ smaller form factor makes them more inconspicuous, allowing them to carry out a range of clandestine tasks including surveillance. It also makes it possible for the UUV to transport numerous robots simultaneously for cluster operations. Adsorption and deformability are features that the claw mechanism’s bionic design gives the robot. The claw mechanism is closed during propulsion to give the robot a streamlined appearance. When the robot gets close to the target, its claws open and its magnetic adsorption unit sticks to the object securely, causing the robot to move in unison with it. During this procedure, the robot uses the reconnaissance module to carry out duties including detection and reconnaissance.
The article’s structure is as follows. The second section provides the general layout of the robot and describes how the Kriging response surface optimization approach was used to improve the magnetic adsorption unit and the claw mechanism, leading to the best possible outcome. We simulated the highest drag force the underwater robot might encounter using Fluent. Afterwards, the lower claws’ opening time for this drag force was determined by dynamic simulation with Adams. By simulating the magnetic adsorption unit with Workbench, the maximum adsorption force was determined. In the third section, we built the robot prototype and used it as the foundation for our pool experiment. We replicated the ejection and secondary propulsion processes of the robot. Measured values of the magnetic adsorption force and the claws opening time were compared with the modeling findings. The fourth section gives some conclusions.

2. Materials and Methods

In this section, the overall structural design of an octopus predation-inspired underwater robot capable of adsorption through opening and closing claws is given. We present a theoretical model of the Kriging response surface optimization method. The claw mechanism’s final scheme was obtained by applying Kriging response surface optimization to find the optimal solutions for the dimensional parameters of the opening and closing claws and the magnetic adsorption unit. Lastly, an analysis and some conclusions are drawn from the simulation of the magnetic adsorption unit and the opening and closing claws.

2.1. Shape and Dimensions

Figure 1 depicts the main layout of the claws opening and closing magnetic adsorption underwater robot. The robot’s diameter is 205 mm, and its body length is 370 mm. The claw mechanism, power module, and propulsion module make up the robot’s primary structural components. The robot’s claw mechanism, which comprises the magnetic adsorption units and the opening and closing claws, is positioned at the front. To maintain the robot’s streamlined form, the claws are closed while it is not absorbed. The robot’s claws open as it gets closer to the target, allowing the magnetic adsorption unit to adhere to the surface and synchronize the robot’s movement with it. The entire robot is powered by the power module, which is located inside the capsule. The robot’s propulsion module is placed in its tail, and its propeller propulsion system allows it to achieve secondary propulsion if it is expelled from the UUV and loses power.

2.2. Claw Mechanism Design

The claw mechanism is made up of two parts: the magnetic adsorption unit and the opening and closing claws. Its specific structure is shown in Figure 2. The claws, magnetic adsorption unit, connecting rod, platform slider, reversing electromagnet, and proximity sensor make up the majority of the mechanism. The opening and closing claws of the robot’s claw mechanism were inspired by the tentacles of an octopus, and the magnetic adsorption unit was based on the suction cups that are present on the tentacles of the same animal. The tension spring elongates into the mechanical energy storage phase before it makes contact with the target, attracted by the reversing electromagnet’s magnetic pull to the metal component at the bottom of the platform slider. The claws close and are incorporated into the housing’s exterior at the same moment. The tension spring in its elongated state drives the platform slider upward, which in turn opens the claws. The magnetic adsorption unit fixed on the surface of the claws touches the target object and adsorbs the entire robot on its surface. All of this happens when the robot gets close to the target when the proximity sensor mounted on top detects the metal and activates the circuit.

2.3. Kriging-Based Response Surface Method

The response surface optimization method, which includes several techniques like experimental design, modeling, determining whether the model is appropriate, and determining the best combination of conditions, is a technique for determining the optimal set of experimental conditions. It is suitable for solving nonlinear data processing and other related problems [15]. By computing the response values of each element, it forecasts the ideal value of the local reaction as well as the test circumstances.
With a fit that is superior to many optimization techniques, the Kriging response surface is a regression algorithm that combines interpolation and Gaussian, and it is generated using both global and local models [16]. Based on the Kriging algorithm, its output parameters are the global design space plus local deviations. It can be expressed as:
y ( x ) = F ( β , x ) + z ( x ) = f T ( x ) β + z ( x ) ,
where  F ( β , x )  is the global model in variable space,  β  is the base function regression coefficient,  x  is the input variable set,  f ( x )  is the polynomial function and is a global model, and  z ( x )  is the Gaussian random function with a mean of 0 and a variance of  2 σ  representing a local error.
The covariance is:
C o v [ Z ( x ) , Z ( u ) ] = σ 2 R ( x , μ , θ ) ,
where  θ  is the parameter of the correlation function, which is used to predict the extent to which the correlation between test points decreases with the distance between the two points; the smaller the correlation, the smoother the generated response surface.
R ( x , μ , θ )  is the variation function between sample points  t  and  u , and it can generally be expressed as:
R ( x , μ , θ ) = j = 1 n R j ( d J ) ,
R j ( d j )  is the kernel function of the variational function and can be expressed as:
R j ( d j ) = j = 1 n e θ j d j 2 ,
where  n  is the dimension of the variational function,  θ  is the parameter of the correlation function, and  d  is the distance between data points.
A correlation matrix on the design sample points is as follows:
R = ( R ( x 1 , x 1 ) R ( x 1 , x n ) R ( x n , x 1 ) R ( x n , x n ) ) ,
where  n  is the total number of data points.
Polynomial parameters of the Kriging model:
β ^ = ( x T R 1 x ) 1 x T R 1 Y ,
where  x  is the matrix of coefficients comprising the design sample and  Y  is the response value corresponding to the design sample.
The estimate of variance is:
σ 2 = ( Y x β ) T R 1 ( Y x β ) ,
Finally, the Kriging response surface is generated with the predicted value at the design point:
y ^ ( x ) = β ^ T f ( x ) + r T ( x ) R 1 ( y F β ^ ) ,
where  r T ( x )  is the relation vector between the unknown and known points.

2.4. Opening and Closing Claws Design

As mentioned in the preceding section, to preserve the robot’s streamlined design, the claws are closed before it makes contact with the target. The robot extends its claws as it gets closer to the target item, and the magnetic adsorption unit on those claws will operate to help the robot stick to the target’s surface. Figure 3’s schematic of the opening and closing claw mechanism was created using the structural makeup of the claw mechanism and its working principle, which are described in Section 2.2 R 1  is the length between the two rotating parts of the claw mechanism and  R 2  is the length of the connecting rod.

2.4.1. Optimized Mathematical Model

Three components make up the mathematical model for the best possible design of opening and closing claws: the constraint variables, design variables, and objective function. The goal function is the angular velocity  ω  at full extension of the claws; the greater its value, the faster the claws open to their full extent. As can be seen from above, the length of the connecting rod and the distance between the two spinning pairs influence the angular velocity when the claws are fully open. As a result, the length  R 1  before the two revolving pairs and the connecting rod’s length  R 2  are chosen as design factors. The range of values for the design variables were determined as  35   mm R 1 45   mm 60   mm R 2 70   mm R 1  and  R 2 ’s changes will affect the overall weight of the claw opening mechanism. The robot needs a lightweight design; the lower the weight, the better; therefore, the mass of the claw opening mechanism  m  is a constraint variable. In summary, the mathematical model for the optimization of the opening and closing claws is obtained as:
{ max ω min m 35   mm R 1 45   mm 60   mm R 2 70   mm ,

2.4.2. Response Surface Establishment

To determine the Kriging response surface, twenty sets of experimental samples were selected, and the extracted data are shown in Table 1.
By fitting a polynomial to the sample points, the response surface seen in Figure 4 was produced. The following conclusions can be drawn from the figure. There is a nearly linear connection between  R 1  and  ω , where  R 1  is smaller and  ω  is greater. The relationship between  R 2  and  ω  is nonlinear; as  R 2  grows,  ω  first increases and then decreases.

2.4.3. Multi-Objective Genetic Algorithm Optimization

More multi-objective optimization is necessary since the mechanism mass  m  was defined by the aforementioned optimization model as a constraint variable. When compared to a traditional genetic algorithm, a multi-objective optimization genetic algorithm solves the objective function using various weight coefficients, improving the objectivity of the optimization process and reducing the complexity of multi-objective problems to single-objective problems [17,18,19]. In addition, the population can be converged towards a Pareto optimum solution using multi-objective genetic algorithms [20].
This paper applies the Workbench multi-objective genetic algorithm module MOGA for additional optimization. A locally optimum solution may be produced more easily with a bigger population, but this comes at the cost of more computation. Therefore, choosing the right population size early in the optimization process is crucial. The total number of initial samples for the design parameters in this paper was 2000, the genetic iterative algorithm selected the samples every 400 generations, the maximum percentage of Parito allowed was set to 70%, and the convergence stability was 2%. The results obtained are shown in Figure 5.
The angular velocity progressively converges to around 18 rad/s as the number of iterations increases, as seen in Figure 5. Consequently, the ideal solution should be approximately 18 rad/s, and the three optimization techniques listed in Table 2 are attained.
Following comparison, it was discovered that there was little variation in the mass change of the three possibilities; hence, option 3, which had the claws’ with a higher angular velocity, was selected. When the parameters were rounded, we got  R 1  is 45 mm,  R 2  is 62 mm.

2.5. Magnetic Adsorption Unit Design

An essential functional element of the claws is the magnetic adsorption unit, which adheres the robot to the target object’s surface when it approaches and maintains a specific level of hardness. Both the magnetic adsorption material and the adsorption structure are important factors affecting the adsorption performance. Materials composed of neodymium, iron, and boron (NbFeB) offer excellent mechanical qualities, a low cost, and very high magnetic induction, coercivity, and maximal magnetic energy product. As a result, it serves as the permanent magnet material in this instance for the magnetic adsorption unit. To create full magnetic circuits and generate stronger magnetic fields, soft magnetic materials serve as bridges in magnetic circuits. Typical examples of soft magnetic metal materials are silicon steel, pure iron, and mild steel. Mild steel is the soft magnetic material of the magnetic adsorption unit in this work because it is less expensive and has a greater magnetization saturation strength than the other materials.
Fixing the permanent magnetic material and building the magnetic circuit are the main goals of the magnetic adsorption unit’s structural design. This paper used an inline construction, where the unthreaded section of the screw was situated in the location shown in Figure 6 by employing a clearance fit between the bolt and the bolt hole. In addition, a soft compression spring was placed under the magnetic block so that it was just slightly compressed by the nut. In this state, it holds the magnetic adsorption unit better. In addition to improving the contact efficiency between the magnet block’s surface and the target surface, this mounting technique fixes the permanent magnet’s difficult-to-fix issue. One claw can be used to install many permanent magnet adsorption devices to provide a suitably high adsorption force. The magnet can make direct contact with the target item thanks to the inline mounting structure, which also eliminates the need for extra pieces that would take up space.
Factors affecting the magnetic attraction  F  are the size of the individual magnets, the distance between the magnets, and the thickness of the magnetic disk. In this paper, a rectangular permanent magnet of equal length and width was used, and the length (width) of the equal length and width rectangular magnetic block is represented by  a , the thickness of the magnetic block is represented by  h 1 , the spacing between the magnetic block and the magnetic block is represented by  L , and the thickness of the magnetic disk is represented by  h 2 , to obtain the mechanical sketch of the magnetic adsorption unit shown in Figure 7.

2.5.1. Optimized Mathematical Model

The goal function and the design variables make up the two components of the magnetically adsorbed unit’s structural optimization model. Due to the tiny range of variation in the magnetic block’s size and its minimal impact on mass, the constraint variables remain unset. The optimization model chooses magnetic attraction as the objective function. The design variables in the optimization model are the thickness of the block, the spacing between blocks, the length (width) of the equal length and width rectangular block, and the thickness of the magnetic disk. The optimization goal is to find the greatest value of the magnetic adsorption force, from which the best solution for each design variable’s dimensions is obtained. Combining the structural requirements of the magnetic adsorption unit and the structural dimensional requirements of the robot, the value ranges of the above four design variables were set as follows: 10   mm a 20   mm 1   mm h 1 5   mm 1   mm L 5   mm 1   mm h 2 5   mm . In conclusion, the following mathematical model for structural optimization of this magnetic adsorption unit is obtained:
{ max F 10   mm a 20   mm 1   mm h 1 5   mm 1   mm L 5   mm 1   mm h 2 5   mm ,

2.5.2. Response Surface Establishment

To determine the Kriging response surface, twenty-five sets of experimental samples were selected, and the extracted data are shown in Table 3.
A polynomial was fitted to the sample points to produce the response surface seen in Figure 8. It is clear from the figure that there is a non-linear relationship between each parameter and drag, and that variations in one design variable have an impact on variations in the drag of the other variables under analysis. Upon examining the distinct response surfaces, it is evident that the objective function’s parameters at the maximum of the magnetic adsorption force are as follows: a = 15   mm h 1 = 5   mm h 2 = 3   mm L = 3   mm . As a result, it is assumed that the aforementioned parameter values represent the optimization model’s ideal solution.

2.6. Simulation of Opening and Closing Claws

The claw mechanism, seen in Figure 9, is derived from the information in Section 2.4. The robot’s front end has four claws, each of which has four grooves for attaching the magnetic blocks. In this section, the robot’s hydrodynamic simulation will be performed using Fluent 2021 R1 software to determine the maximum drag force it experiences. Adams 2018 software will then model and simulate the claws’ dynamics to determine the claws opening time.

2.6.1. Hydrodynamic Forces Estimation

First and foremost, excellent grid division, watershed modeling, and other pre-processing tasks were performed. Next, the boundary conditions were finished by utilizing the Name selection tool to establish the watershed’s inflow and outflow boundaries, as seen in Figure 10. Once the Fluent program was opened, liquid water was chosen as the fluid medium, the viscosity setting was adjusted to the k-epsilon model, and the boundary setting was adjusted to include the 0.5 m/s velocity needed for secondary propulsion. The default solution method was chosen, the convergence curve and residual value in the method was adjusted, the drag computation drag was added to the solver, and solving after initialization began to obtain the convergence curve as seen in Figure 11. It should be noted that the hydrodynamic parameters required for simulation calculation in this paper were selected by experience. In order to verify the accuracy of the results, we conducted a water drag experiment on the robot, and the results obtained by the experiment were not much different from the simulation results.
The post-processing module was entered after receiving the convergence curve as seen in Figure 11. Figure 12 illustrates the successive creation of the pressure distribution cloud, velocity vector map, and pressure distribution cloud of the four magnetically adsorbed grips at full opening.
From Figure 12, it can be found that the robot’s head was subjected to the greatest compressive force, the drag of the robot as a whole was 7.526 N, and the drag of the claws was 1.238 N. This drag represents the claws’ highest drag since the magnetically adsorbed claws have the biggest flow area when they are opening. The water drag experiment was carried out to verify the correctness of the water drag simulation results. The robot was fixed in the center of the circulation tank using a suspension robot arm. The six-component balance was fixed between the robot and the suspension robot arm to collect water drag. The flow controller was used to set the flow speed to 0.5 m/s. Finally, the drag borne by the robot was about 7.771 N, which is not much different from the simulation result of 7.526 N.

2.6.2. Dynamic Simulation

The claw mechanism was reduced to an analogous model, as seen in Figure 13 (ignoring gravity and assuming the robot is in a zero-buoyancy condition). Each of the parts were then examined independently for forces.
The equilibrium equations were provided and the forces acting on member 1 were analyzed. Equation (11) provides the equilibrium equation, and Figure 14 displays the force results.
{ F 21 F 41 F W = 0 F 21 ( S 1 y B y ) = J 1 φ ¨ 1 ,
where  F 21  is the restraining reaction force on member 1 from member 2, N;  F 41  is the restraining reaction force on member 1 from member 4, N;  F W  is the water drag of member 1, N;  S 1 y  is the value of the y-axis coordinate of the center of mass of member 1;  B y  is the y-axis coordinate value of point B on member 1;  J 1  is the moment of inertia of member 1,  kg m φ ¨ 1  is the angular acceleration of member 1, rad/s2.
The forces acting on member 2 were identified and equations for equilibrium were created. Equation (12) provides the equilibrium equation, and the force results are displayed in Figure 15.
{ F 32 F 12 = 0 F 12 ( B y S 2 y ) F 32 ( S 2 y C y ) = J 2 φ ¨ 2 ,
where  F 12  is the restraining reaction force on member 2 from member 1, N;  F 32  is the restraining reaction force on member 2 from member 3, N;  S 2 y  is the value of the y-axis coordinate of the center of mass of member 2;  C y  is the y-axis coordinate value of point C on member 1;  J 2  is the moment of inertia of member 2,  kg m φ ¨ 2  is the angular acceleration of member 2, rad/s2.
The equilibrium equations were provided and the forces acting on member 3 were analyzed. Equation (13) provides the equilibrium equation, and Figure 16 displays the force results.
F s p F 23 x = m 3 x ¨ ,
where  F s p  is the spring tension on member 3, N;  F 23  is the restraining reaction force on member 3 from member 2, N;  F i 3  is the inertial force on member 3, N.
After importing the claw mechanism from the previous section into Adams, three new vices were added: a rotating vice between the claws and the connecting rod and other components, a fixed vice between the center rod and the earth, and a moving vice between the platform slider and the center rod. The creation of the simulation animation was performed once all the loads were configured by the force analysis mentioned above. It is possible to successfully open the claws and finish the adsorption operation by entering the simulation module, running the interactive simulation, and adjusting the length and step size to see the simulation animation. The outcomes are displayed in Figure 17.
The platform slider caused a Y-axis displacement of 60 mm from the starting position to the claws’ full opening, according to measurements taken from the Solidworks model. Thus, in the Adams model, the time needed for the claws to completely open was equivalent to the time needed for the platform slider component to ascend 60 mm. As seen in Figure 18, the Y-axis translation displacement of the platform slider was chosen, and a displacement-time curve was produced. After importing the image into the post-processor, the corresponding time was recorded when the Y-axis displacement difference reached 60 mm. In this case, the Y-axis starting position was approximately Y = −0.065, which is equivalent to the value of X at the intersection of the image and the Y = −0.005 straight line, with a time interval of approximately 0.1 s.
In conclusion, the claw mechanism may theoretically open quickly enough to open normally even in the presence of the highest water flow drag, and the adsorption action was completed in 0.1 s.

2.7. Simulation of Magnetic Adsorption Unit

Section 2.5 allows for the determination of the following magnetic adsorption unit layout. Four permanent magnet neodymium-iron-boron (NdFeB) blocks were bolted into each claw of the magnetic adsorption unit via bolt-in installation. Each block had a thin layer of mild steel magnet-conducting sheet attached to the bottom of it. The magnetic blocks were rectangular blocks measuring 15 × 15 × 5 mm, with equal width and length. The magnetic guide sheet had a thickness of 3 mm and was positioned four blocks apart equally. We bolted on the magnetic block, claw, and magnetic disk. The magnetic adsorption unit mentioned above was reduced in complexity to the form seen in Figure 19, and an iron plate with a thickness of 10 mm was included to replicate the adsorption target.
After that, the model was loaded into the Ansys workbench platform’s Magnetostatic module. We used the DM geometric model modification module to add air contours, i.e., boundary conditions, and create a magnetized coordinate system with material properties and assigned them to the entities. The model after meshing is displayed in Figure 20. The magnet block in this model was meshed with a 1 mm hexahedral mesh, while the remaining entities were meshed with a 5 mm mesh.
The matching goal results were entered into the Solution module once the grid had been split, and they were then computed to create the magnetic density cloud map, as seen in Figure 21. The magnetic adsorption force found was 120.008 N, and it is evident that the magnetic density streamlines were highly dense. A magnetic adsorption force of around 480 N may be produced by the four magnetic adsorption grips.

3. Results and Discussion

As stated in the preceding section, we used 3D printing technology to develop a test prototype, depicted in Figure 22, of a claw opening and closing magnetic adsorption underwater robot. In this paper, DSM’s Somos®14120 was used as a 3D printing consumable. Somos®14120 is a low-viscosity liquid photosensitive polymer that produces strong, tough, and waterproof parts. Parts made with Somos®14120 have a white, opaque appearance, similar to plastics used in production. Somos®14120 has many properties that mimic traditional engineering plastics, including ABS and PBT. The bending strength of the material is 68.9 MPa, the hardness is 81 HD, and the water absorption is 0.24%. To confirm the robot’s viability, this part was carried pool experiments using the test prototype. These studies included robotic launching, claws opening, and magnetic adsorption force tests. The robot’s viability was confirmed by the experimental findings.

3.1. Robot Ejection Experiment

The launching platform, a high-speed camera image acquisition system, and a robot made up the robotic underwater ejection experiment system. There were three primary stages to the ejection experiment. The first stage utilizes the launch platform’s pneumatic launch tubes to eject the robot towards the target. Using a propeller at its end, the robot’s second stage uses secondary propulsion to get closer to the target. In the third stage the robot’s claws are opened to attach to the target surface using the adsorption force of the magnetic adsorption unit. To rapidly transfer the underwater robot to the target location, this research used a pneumatic launch, in which the underwater robot received its initial power from the launch platform. The robot is forced to travel inside the launcher tube and finally eject out of it by applying compressed gas from the large-capacity gas cylinder at the end of the pneumatic launcher tube to the robot’s tail. During the launching procedure, the robot’s motion status was recorded by a high-speed camera. The high-speed camera model VE0710 L used in this experiment can achieve a shooting speed of 680,000 photos/SEC at the lowest pixel (64*8 pixels) and 7500 photos/SEC at 1280*800 pixels. It can be used with ISO10000 color and ISO32000 black and white cameras to achieve high-definition video capture in extreme shock, vibration, and temperature environments. In addition, the high-speed camera was put into a sealed cylinder to waterproof it. The robot was put through a pool launching experiment once the experimental apparatus was installed; the results are displayed in Figure 23.
Numerical calculation software was used to process the data from the high-number camera acquisition system to determine the hydrodynamic parameters of the robot during launch. The velocity versus time graph for the robot’s launch out of the cylinder procedure is displayed in Figure 24. The robot was ejected from the pneumatic launcher tube with almost no speed change before 0.1 s, as shown in the figure. The robot’s speed increased gradually after 0.1 s and reached its peak at 0.3 s. After that, the robot travelled a predetermined distance at 7.5 m/s before the propeller at its tail end began to operate to realize secondary propulsion, which ultimately propelled the robot to the target.

3.2. Claws Opening Experiment

An underwater high-speed camera with a transparent sealed waterproof chamber was used in the experiments to capture and record the speed of the robot’s head each time it reached the target. This allowed researchers to measure the robot’s speed to reach the target as well as the time needed to open the claws and complete the adsorption maneuver. As indicated in Table 4, the gathered data were entered into the computer and arranged into 20 sets of experimental data, demonstrating that the robot was effective in absorbing the targets in each trial. With an average time of around 0.11 s, the claws all finished the adsorption activity before the robot’s head reached the target. This time was quite similar to the simulation data of 0.1 s found in Section 2.6. It demonstrated that, while sprinting toward the target, the robot can swiftly open its claws and finish the adsorption process on the target surface.

3.3. Magnetic Adsorption Experiment

The adsorption force of the magnetic adsorption unit was obtained by dragging the robot using a wide range of dynamometers after each launch adsorption on a steel plate, observing the dynamometer indices, and noting the experimental data displayed by the dynamometers when the robot was detached from the iron plate. This allows us to simulate the disturbance force of the robot in the water caused by the lateral and longitudinal currents. Figure 25 provides an overview of the experimental process.
The dynamometer signals were sorted into the data that Table 5 displays. The NdFeB-based magnetic adsorption unit had strong adsorption capabilities and the ability to stabilize the robot on the target iron plate. The actual observed magnetic adsorption force was less than the theoretically calculated magnetic adsorption force in Section 2.7 because of experimental imperfections. In the longitudinal direction, the robot can achieve a magnetic adsorption force of 304.73 N, which is more securely adsorbed, and the real measured robot can obtain an average lateral magnetic adsorption force of 462.42 N, which is around 18 N different from the simulation data of 480 N.

3.4. Limitations and Prospects

To sum up, we obtained the required data through the pool experiment. By comparing the experimental results with the simulation results in Section 2, it is not difficult to find that the robot prototype can quickly open the claws and carry out stable and firm adsorption on the target object. Although we have successfully verified the feasibility of this robot, it still has some limitations due to various reasons. Next, we will discuss its limitations and propose some prospects for future work. First of all, because the adsorption structure of the robot are the magnetic adsorption units, the robot can only be adsorbed on the magnetic surface, which has certain limitations. In addition, the robot is made using 3D printing, and its material properties and structural strength need to be improved. Not only that, although the smaller shape and size improves the invisibility of the robot, it also makes the energy it can carry relatively small, and the endurance is relatively poor.
In this paper, there is room for optimization in the design of the adsorption unit. In the future, we will study and design magnetic adsorption units with stronger adsorption force, or adopt more diversified adsorption methods to adapt to more adsorption target materials. Since the robot was made by 3D printing, its material structure strength is poor. In the future, we will study the use of materials with better performance to make the robot to obtain better structural performance.

4. Conclusions

This work describes the creation of an octopus predation-inspired underwater robot capable of adsorption through opening and closing claws. The three primary components of the robot are the propeller propulsion module at the tail end, the power module within the capsule, and the claw mechanism at the front end. Mission payload devices, such as reconnaissance modules, can be mounted within the capsule to enable underwater missions. The opening and closing claws that mimic an octopus tentacle were created. It shuts while the robot is in a non-absorption stage, which is when it keeps its overall streamlined form. By mimicking how suction cups on an octopus’s tentacles adsorb prey while feeding, a magnetic adsorption unit was created. The robot’s claw mechanism opens as it approaches the target, completing the adsorption process on the target surface with the help of its magnetic adsorption unit. The robot’s claw mechanism is made up of the magnetic adsorption unit and opening and closing claws. An underwater unmanned vehicle carries and transports the robot to a pre-designated location, from whence it uses the launcher to propel itself toward the target. The robot then extends its claws to connect to the target after being driven a certain distance by its power unit, the propulsion module. Following the completion of adsorption, the robot travels in lockstep with the target, enabling the functional modules installed on it to perform duties like detection and reconnaissance. Below is a list of the work completed in this paper:
  • First, the robot’s general structural design is provided, and the particular structural design of the claw mechanism—that is, the magnetic adsorption unit and the opening and closing claws—is finished. The primary design factors were determined after providing physical models of the magnetic adsorption unit and the opening and closing claws, respectively. Using the Kriging response surface optimization approach, the claw mechanism’s design parameters were optimally solved, leading to the final production of the robot’s overall structural model;
  • We finished simulating the hand–jaw system theoretically. Using Fluent 2021 R1software, the robot’s maximum drag was measured, and the claw mechanism’s dynamics model was created. The dynamic simulation of the claw mechanism was finished by importing the aforementioned drag and the dynamics model’s boundary conditions into Adams 2018 software. This allows for the determination of the opening and closing claws’ opening time. The magnetostatic module in the Workbench program was used to perform the finite element simulation of the magnetically adsorbed nano element to determine the maximum adsorption force that the magnetically adsorbed unit is capable of providing;
  • To confirm the robot’s viability, a test prototype was constructed and pool experiments were carried out. An HD camera captured the testing data while a pneumatic launching tube replicated the robot’s ejection mechanism. The feasibility of the robot’s functions was confirmed by testing the launch, secondary propulsion, and adsorption phases. The high-speed camera captured the robot’s claws opening time. The robot’s ability to swiftly expand its claws to finish the adsorption on the target surface when it was reached was confirmed by the experimental findings, which closely resemble the modeling results. A dynamometer was used to record the robot’s lateral and longitudinal adsorption forces as it adhered to the target surface. The robot’s ability to firmly adsorb on the target surface and the little discrepancy between the numerical values and the simulation results confirmed that the robot’s magnetic adsorption unit can provide an extremely steady adsorption force.

Author Contributions

Conceptualization, H.G.; Methodology, F.M.; Software, F.M.; Formal analysis, B.T.; Resources, S.H.; Writing—original draft, Z.L. and B.T.; Writing—review & editing, Z.L.; Visualization, H.G.; Supervision, S.H.; Project administration, H.G. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by Key R&D Program of Shandong Province of China, grant number 2021JMR0302, and the National Natural Science Foundation of China, grant number 5207110396.

Data Availability Statement

The raw data supporting the conclusions of this article will be made available by the authors on request.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Hall, J.K. GEBCO Centennial Special Issue–Charting the secret world of the ocean floor: The GEBCO project 1903–2003. Mar. Geophys. Res. 2006, 27, 1–5. [Google Scholar] [CrossRef]
  2. Christianson, C.; Bayag, C.; Li, G.; Jadhav, S.; Giri, A.; Agba, C.; Li, T.; Tolley, M. Jellyfish-Inspired Soft Robot Driven by Fluid Electrode Dielectric Organic Robotic Actuators. Front. Robot. AI 2019, 6, 126. [Google Scholar] [CrossRef] [PubMed]
  3. Katzschmann, R.K.; DelPreto, J.; MacCurdy, R.; Rus, D. Exploration of underwater life with an acoustically controlled soft robotic fish. Sci. Robot. 2018, 3, eaar3449. [Google Scholar] [CrossRef] [PubMed]
  4. Russo, R.S.; Blemker, S.S.; Fish, F.E.; Bart-Smith, H. Biomechanical model of batoid (skates and rays) pectoral fins predicts the influence of skeletal structure on fin kinematics: Implications for bio-inspired design. Bioinspir. Biomim. 2015, 10, 046002. [Google Scholar] [CrossRef]
  5. Xing, C.; Cao, Y.; Cao, Y.; Pan, G.; Huang, Q. Asymmetrical Oscillating Morphology Hydrodynamic Performance of a Novel Bionic Pectoral Fin. J. Mar. Sci. Eng. 2022, 10, 289. [Google Scholar] [CrossRef]
  6. Li, G.; Liu, G.; Leng, D.; Fang, X.; Li, G.; Wang, W. Underwater Undulating Propulsion Biomimetic Robots: A Review. Biomimetics 2023, 8, 318. [Google Scholar] [CrossRef] [PubMed]
  7. Li, Y.; Xu, Y.; Wu, Z.; Ma, L.; Guo, M.; Li, Z.; Li, Y. A comprehensive review on fish-inspired robots. Int. J. Adv. Robot. Syst. 2022, 19, 17298806221103707. [Google Scholar] [CrossRef]
  8. Sun, B.; Li, W.; Wang, Z.; Zhu, Y.; He, Q.; Guan, X.; Dai, G.; Yuan, D.; Li, A.; Cui, W.; et al. Recent Progress in Modeling and Control of Bio-Inspired Fish Robots. J. Mar. Sci. Eng. 2022, 10, 773. [Google Scholar] [CrossRef]
  9. Qu, J.; Xu, Y.; Li, Z.; Yu, Z.; Mao, B.; Wang, Y.; Wang, Z.; Fan, Q.; Qian, X.; Zhang, M.; et al. Recent Advances on Underwater Soft Robots. Adv. Intell. Syst. 2023, 6, 2300299. [Google Scholar] [CrossRef]
  10. Youssef, S.M.; Soliman, M.; Saleh, M.A.; Mousa, M.A.; Elsamanty, M.; Radwan, A.G. Underwater Soft Robotics: A Review of Bioinspiration in Design, Actuation, Modeling, and Control. Micromachines 2022, 13, 110. [Google Scholar] [CrossRef] [PubMed]
  11. Giorgio-Serchi, F.; Arienti, A.; Corucci, F.; Giorelli, M.; Laschi, C. Hybrid parameter identification of a multi-modal underwater soft robot. Bioinspir. Biomim. 2017, 12, 025007. [Google Scholar] [CrossRef] [PubMed]
  12. Renda, F.; Giorgio-Serchi, F.; Boyer, F.; Laschi, C.; Dias, J.; Seneviratne, L. A unified multi-soft-body dynamic model for underwater soft robots. Int. J. Robot. Res. 2018, 37, 648–666. [Google Scholar] [CrossRef]
  13. Cianchetti, M.; Calisti, M.; Margheri, L.; Kuba, M.; Laschi, C. Bioinspired locomotion and grasping in water: The soft eight-arm OCTOPUS robot. Bioinspir. Biomim. 2015, 10, 035003. [Google Scholar] [CrossRef] [PubMed]
  14. Calisti, M.; Giorelli, M.; Levy, G.; Mazzolai, B.; Hochner, B.; Laschi, C.; Dario, P. An octopus-bioinspired solution to movement and manipulation for soft robots. Bioinspir. Biomim. 2011, 6, 036002. [Google Scholar] [CrossRef] [PubMed]
  15. Xu, X.; Hu, X.; Wu, S.; Zhao, Q. Experimental Design and Response Surface of Kriging in the Application of the Optimization Design. Modul. Mach. Tool Autom. Manuf. Tech. 2017, 43, 62–65+69. [Google Scholar] [CrossRef]
  16. Mifsud, D.; Verdin, P.G. Surrogate-Based Design Optimisation Tool for Dual-Phase Fluid Driving Jet Pump Apparatus. Arch. Comput. Methods Eng. 2021, 28, 53–89. [Google Scholar] [CrossRef]
  17. Deb, K. Multi-objective genetic algorithms: Problem difficulties and construction of test problems. Evol. Comput. 1999, 7, 205–230. [Google Scholar] [CrossRef] [PubMed]
  18. Li, L.; Pan, F. Survey of multi-objective evolutionary algorithm based on genetic algorithm. In Proceedings of the Chinese Control Decision Conference, Wuxi, China, 3–6 July 2007; pp. 363–366. [Google Scholar]
  19. Zhang, S.; Wang, H.; Yang, D.; Huang, M. Hybrid Multi-Objective Genetic Algorithm for Multi-Objective Optimization Problems. In Proceedings of the IEEE 27th Chinese Control and Decision Conference (CCDC), Qingdao, China, 23–25 May 2015; pp. 1970–1974. [Google Scholar]
  20. Long, Q.; Wu, C.; Huang, T.; Wang, X. A genetic algorithm for unconstrained multi-objective optimization. Swarm Evol. Comput. 2015, 22, 1–14. [Google Scholar] [CrossRef]
Figure 1. General structural model of the robot.
Figure 1. General structural model of the robot.
Applsci 14 02250 g001
Figure 2. Claw mechanism model.
Figure 2. Claw mechanism model.
Applsci 14 02250 g002
Figure 3. Sketch of claw mechanism.
Figure 3. Sketch of claw mechanism.
Applsci 14 02250 g003
Figure 4. Response surface of  R 1 R 2 .
Figure 4. Response surface of  R 1 R 2 .
Applsci 14 02250 g004
Figure 5. Variation of angular velocity with number of iterations.
Figure 5. Variation of angular velocity with number of iterations.
Applsci 14 02250 g005
Figure 6. Schematic structure of magnetic adsorption unit.
Figure 6. Schematic structure of magnetic adsorption unit.
Applsci 14 02250 g006
Figure 7. Magnetic adsorption unit mechanism sketch.
Figure 7. Magnetic adsorption unit mechanism sketch.
Applsci 14 02250 g007
Figure 8. Magnetic attraction about variable interaction affecting the response surface: (a a h 1 ; (b a L ; (c a h 2 ; (d L h 1 ; (e L h 2 ; (f h 1 h 2 .
Figure 8. Magnetic attraction about variable interaction affecting the response surface: (a a h 1 ; (b a L ; (c a h 2 ; (d L h 1 ; (e L h 2 ; (f h 1 h 2 .
Applsci 14 02250 g008
Figure 9. Opening and closing claws: (a) Closing state; (b) Opening state.
Figure 9. Opening and closing claws: (a) Closing state; (b) Opening state.
Applsci 14 02250 g009
Figure 10. Pretreatment stage.
Figure 10. Pretreatment stage.
Applsci 14 02250 g010
Figure 11. Convergence curve.
Figure 11. Convergence curve.
Applsci 14 02250 g011
Figure 12. After treatment stage: (a) pressure distribution of the robot; (b) velocity vector cloud graphics; (c) claws pressure distribution.
Figure 12. After treatment stage: (a) pressure distribution of the robot; (b) velocity vector cloud graphics; (c) claws pressure distribution.
Applsci 14 02250 g012
Figure 13. Simplified model of claw mechanism.
Figure 13. Simplified model of claw mechanism.
Applsci 14 02250 g013
Figure 14. Schematic of force on member 1.
Figure 14. Schematic of force on member 1.
Applsci 14 02250 g014
Figure 15. Schematic of force on member 2.
Figure 15. Schematic of force on member 2.
Applsci 14 02250 g015
Figure 16. Schematic of force on member 3.
Figure 16. Schematic of force on member 3.
Applsci 14 02250 g016
Figure 17. Screenshot of simulation animation: (a) before the claws open; (b) claws opening; (c) claws opening complete.
Figure 17. Screenshot of simulation animation: (a) before the claws open; (b) claws opening; (c) claws opening complete.
Applsci 14 02250 g017
Figure 18. Displacement-time curve of the platform slider.
Figure 18. Displacement-time curve of the platform slider.
Applsci 14 02250 g018
Figure 19. Simplified model of magnetic adsorption.
Figure 19. Simplified model of magnetic adsorption.
Applsci 14 02250 g019
Figure 20. Meshing.
Figure 20. Meshing.
Applsci 14 02250 g020
Figure 21. Magnetic density streamline.
Figure 21. Magnetic density streamline.
Applsci 14 02250 g021
Figure 22. The robot test prototype: (a) claws closing; (b) claws opening.
Figure 22. The robot test prototype: (a) claws closing; (b) claws opening.
Applsci 14 02250 g022
Figure 23. Robot ejection experiment: (a) ejection phase; (b) secondary advancement phase; (c) adsorption phase.
Figure 23. Robot ejection experiment: (a) ejection phase; (b) secondary advancement phase; (c) adsorption phase.
Applsci 14 02250 g023
Figure 24. Velocity–time curve.
Figure 24. Velocity–time curve.
Applsci 14 02250 g024
Figure 25. Magnetic adsorption experiment: (a) longitudinal force application; (b) lateral force application.
Figure 25. Magnetic adsorption experiment: (a) longitudinal force application; (b) lateral force application.
Applsci 14 02250 g025
Table 1. Experimental sample data of claws.
Table 1. Experimental sample data of claws.
NumberR1 (mm)R2 (mm)ω (mm)m (kg)
140.2562.2517.2350.486
242.2566.7517.6480.489
335.2568.7516.2850.490
441.2560.2517.4010.485
536.2564.7516.4660.488
644.7563.7518.0720.487
740.7565.7517.3620.488
844.2564.2517.9880.487
938.2569.2516.9010.490
1037.7561.2516.7400.486
1139.2565.2517.0700.488
1237.2566.2516.6790.489
1339.7562.7517.1440.487
1442.7567.2517.7460.489
1538.7561.7516.9380.486
1643.7560.7517.8590.485
1743.2568.2517.8480.490
1841.7567.7517.5620.489
1936.7563.2516.5570.487
2035.7569.7516.3950.491
Table 2. Optimization solutions.
Table 2. Optimization solutions.
NumberR1 (mm)R2 (mm)ω (mm)m (kg)
144.9996018.070.485
244.99861.02818.0830.486
344.99662.02818.0950.487
Table 3. Experimental sample data of magnetic adsorption units.
Table 3. Experimental sample data of magnetic adsorption units.
NumberL (mm)a (mm)h1 (mm)h2 (mm)F (N)
1312.53371.079
21103337.805
35103343.703
41153399.704
551533108.295
6312.51118.769
7312.55169.471
8312.51529.303
9312.55596.753
10112.53147.161
11512.53149.076
12112.53572.217
13512.53580.746
143101317.401
153151341.816
163105350.822
1731553120.008
18112.51327.259
19512.51329.191
20112.55381.744
21512.55392.839
223103131.073
233153168.994
243103544.849
2531535118.971
Table 4. Experimental data of claws opening.
Table 4. Experimental data of claws opening.
NumberVelocity (m/s)Yes or NoTime (s)
10.32Yes0.11
20.31Yes0.12
30.32Yes0.11
40.33Yes0.10
50.37Yes0.11
60.40Yes0.10
70.36Yes0.12
80.37Yes0.10
90.39Yes0.09
100.40Yes0.10
110.41Yes0.13
120.44Yes0.12
130.42Yes0.09
140.43Yes0.10
150.51Yes0.10
160.48Yes0.09
170.49Yes0.11
180.53Yes0.12
190.49Yes0.10
200.50Yes0.11
Table 5. Experimental data of magnetic adsorption.
Table 5. Experimental data of magnetic adsorption.
NumberTransverse Force (N)Longitudinal Force (N)
1457.31305.21
2464.30306.14
3458.09316.82
4473.27304.98
5464.40298.46
6453.18307.45
7468.67309.06
8469.14304.25
9454.65294.08
10461.15300.80
average 462.42304.73
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

Gong, H.; Li, Z.; Meng, F.; Tan, B.; Hou, S. Octopus Predation-Inspired Underwater Robot Capable of Adsorption through Opening and Closing Claws. Appl. Sci. 2024, 14, 2250. https://doi.org/10.3390/app14062250

AMA Style

Gong H, Li Z, Meng F, Tan B, Hou S. Octopus Predation-Inspired Underwater Robot Capable of Adsorption through Opening and Closing Claws. Applied Sciences. 2024; 14(6):2250. https://doi.org/10.3390/app14062250

Chicago/Turabian Style

Gong, Haixia, Zicong Li, Fance Meng, Bowen Tan, and Shuping Hou. 2024. "Octopus Predation-Inspired Underwater Robot Capable of Adsorption through Opening and Closing Claws" Applied Sciences 14, no. 6: 2250. https://doi.org/10.3390/app14062250

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop