2.1. Structural design of robot cell module

Aiming at the problem that the traditional manipulator has a single function and limited operating range, which cannot better match the future on-orbit assembly tasks, this section designs four kinds of cell modules based on the self-reconfigurable modular theory. The truss cell robot adopts the idea of bionics, and the most basic unit of the robot is called the cell. In the microgravity space environment, the cellular robot is connected to a robot that can realize complex actions by combining different cellular modules, which is called reconfiguration. Aiming at the on-orbit assembly task of space trusses with different requirements and complex environments, researchers can use different reconstruction strategies to flexibly complete the corresponding work objectives. A truss cell robot is a complex machine consisting of diverse cell modules that carry out specific and unique functions. These cell modules work together in harmony to achieve the overall goal of the robot, namely to navigate and operate effectively on trusses and other complex structures. In this paper, cell modules are divided into four categories: one is the interstitial cell module that plays the role of connection and support, and the other three are the functional cell modules that can perform the most basic actions, namely, the 180 swing cell module, the 360 rotation cell module, and the gripper cell module. The shapes of these four cell modules are all cubes with a side length of 200 mm.

As shown in Fig. 1, there are four kinds of cell modules. The whole connecting mechanism of the interstitial cell module can be divided into the external butting surface part and the internal mechanical transmission part, and the internal mechanical transmission part can be divided into translation movement and rotation movement according to the movement mode. Its working principle is that the internal motor controls the transmission part to make the clamping block assembly extend outward from the connection surface through linear translation motion, and at the same time, the clamping block assembly makes radial expansion motion along the bevel gear through rotary motion. The combination of the two motions makes the clamping block assembly firmly stuck in the clamping groove of the external docking surface and finally realizes the bidirectional connection function of interstitial cells. The connecting mechanism used in this design occupies a small space, and each interstitial cell module can have two sets of active connecting mechanisms; that is, the internal space layout of the module is reasonable so that the two sets of active connecting mechanisms can work at the same time and do not interfere with each other, further improving the docking efficiency of interstitial cells. Moreover, the current design utilizes a mechanical slot connection approach, which offers numerous benefits. One of the major advantages of this method is its high connection stability, ensuring a secure and reliable connection. Additionally, the slot connection design features strong connection strength and can withstand high levels of mechanical stress and heavy loads. In general, the mechanical slot connection design employed in this design provides exceptional durability and robustness, making it an ideal choice for a range of applications. The swing cell module consists of two rotating shells. The main components installed in the upper rotating shell are the motor, gear, worm gear, and rotating shaft. The power is output from the motor, transmitted to the worm gear through the gear and finally transmitted to the worm gear shaft. The lower shell is connected with the upper rotating shaft, and the power of the upper rotating shell can be transmitted to the lower rotating shell through the rotating shaft of the worm wheel, so as to realize the swinging movement of 180°. The worm wheel and worm can make the cell module mechanism realize the self-locking function, which is more effective and stable in the rotation process. Each component is fixed on the rotator, and when the swing cell module rotates, the internal components can operate stably and effectively without being affected by the rotation. The internal structure of the 360° rotating cell module consists of an upper rotating shell and a lower rotating shell, and its working principle is similar to that of the swinging cell module. The worm gear transmission mode is adopted, and the power of the upper rotating body is output through the rotating shaft, and the lower rotating body is connected with the rotating shaft through the connecting disk to realize the rotation of the lower rotating body. The motor shaft in the rotating module is connected with the hole on the worm, and the worm is directly connected with the motor shaft through an interference fit to drive the worm to rotate, thus eliminating the gear transmission part in the swing cell module. Functionally, the advantage of the rotating cell module is that it can rotate 360°, which increases the freedom of rotation. The worm gear structure has stable transmission and good self-locking performance, which can stop the lower rotating shell from rotating at any position. The gripper cell module can be used in many aspects, such as gripping objects, building trusses, transporting trusses, and so on. In the robot design in this paper, the clamping function is to meet the needs of the truss climbing task of an on-orbit truss cell robot. The linear movement of the lead screw nut is caused by the rotational motion of the servo motor, which drives the lead screw. Subsequently, the connecting parts transmit this linear motion to the short arm of the gripper, enabling it to move vertically, so that the long arm can be driven to open and close, so that the whole gripper has two actions of clamping and opening, and the truss can be clamped and loosened. The servo motor can control the output torque more accurately, and it is more conducive to the digital control of the gripper cell module.

Figure 1. Structure diagram of cell unit module.

2.2. Configuration design of cellular robot

In this section, the serial configuration of cellular robots is designed. The requirements for a reconstructed cellular robot undertaking the on-orbit assembly task of space trusses include various abilities. Firstly, the robot must be capable of performing stable grasping, transporting, and assembling actions on the truss components. Additionally, the robot should be able to move smoothly from its initial position to the designated target location with high precision and reliability. The above-mentioned abilities are imperative to ensure that the robot can execute its tasks effectively in a challenging and complex space environment, where accuracy, stability, and safety are of utmost importance. Achieving these requirements will enhance the robot’s performance and flexibility, opening up new possibilities for space truss handling and on-orbit assembly. This requires the reconstructed robot to have enough freedom to support it to complete various grasping and climbing actions in the process of performing tasks.

In the space environment, the cellular robot needs at least 3 degrees of freedom to climb and walk on the truss. Moreover, because the end effector needs one swing degree of freedom to adjust the posture when the cellular robot carries out the truss installation task, the cellular robot needs at least four degrees of freedom to ensure the ability to assemble the truss after reconstruction, including two rotational degrees of freedom at both ends and two swing degrees of freedom in the middle. The number of degrees of freedom of the robot should be suitable for its configuration. Too many degrees of freedom, such as a seven-degree-of-freedom redundant cell robot, will waste the resources of cell modules.

Therefore, based on the perspective of human bionics, this paper designs a six-degree-of-freedom cell robot, as shown in Fig. 2. The robot configuration is similar to the human body, and the first joint uses a rotating cell module to output rotating action, which is similar to the human core. The second and third joints use swing cell modules to output swing action, which is similar to human big and small arms. The fourth and fifth joints use swing cell modules, and the sixth joint uses rotating cell modules. These three joints are similar to human wrists. This configuration is not limited by the rotation angle of the joint module and can assemble truss rods in any position and posture in the space truss environment, which ensures that the robot can stably and flexibly accomplish the target task on the truss and has strong on-orbit assembly efficiency.

Figure 2. Six-DOF in-orbit truss cell robot.

3.1. Improved sparrow search algorithm design

Within the SSA algorithm, the discoverer plays a crucial role in determining the foraging direction for the entire population. Consequently, it is essential for the discoverer to possess a higher fitness value and a broader search range. Additionally, the initial position of the individual population significantly impacts the optimization performance of the algorithm. To enhance global exploration capabilities during the update of the discoverer’s position and mitigate the negative impact of local optimal solutions in later iterations, it introduces chaotic mapping for the initialization of the sparrow population. The update process for the discoverer’s location is described as follows:

(1) \begin{equation} X_{i,j}^{t+1}=\left\{\begin{array}{l@{\quad}l} X_{i,j}^{t}\cdot \exp \!\left(-\dfrac{i}{\alpha \cdot \text{iter}_{\max}}\right) & \text{ if }\quad R_{2}\lt \text{ST} \\[12pt] X_{i,j}^{t}+Q\cdot L & \text{ if }\quad R_{2}\geq \text{ST} \end{array}\right. \end{equation}

In the given context, the variables and notations are described as follows: t denotes the current iteration number, and j takes integer values from 1 to d representing the dimension information, where d signifies the dimensionality. $\text{iter}_{\max}$ is a constant that signifies the maximum number of iterations. $X_{ij}$ denotes the position information of the ith sparrow in the jth dimension. α represents a random number sampled from the uniform distribution in the interval $(0,1]$ . $R_{2}(R_{2}\in [0,1])$ and $\text{ST}(\text{ST}\in [0.5,1])$ represent the early warning value and safety value, respectively. Q is a random number following a normal distribution. L denotes a 1×d matrix, where all elements in the matrix are equal to 1.

To ensure a random and regular distribution of the sparrow population, it is necessary to apply a chaos operator that maps the initial sparrow population. This ensures that the sparrow individuals traverse all positions within the search range. The comparative analysis shown in Fig. 3 demonstrates the superior distribution characteristics of the cubic map compared to the classical logistic map. Notably, the cubic map significantly enhances the diversity of the sparrow algorithm, especially in the later stages. Based on this observation, we include the chaotic operator from the cubic map in the SSA, aiming to maximize global search efficiency.

Figure 3. Distribution histogram of chaotic mapping.

Chaos operator has the advantages of randomness and regularity, and the cubic chaotic map has better uniform distribution performance between [0,1]. Chaos operator has the advantages of randomness and regularity, and the cubic chaotic map has better uniform distribution performance between [0,1]. The formula for the cubic mapping is defined as follows:

(2) \begin{equation} \begin{array}{l} y\left(i+1\right)=4y\left(i\right)^{3}-3y\left(i\right)\\[5pt] -1\lt y\left(i\right)\lt 1,y\left(i\right)\neq 0,i=0,1,\ldots,n \end{array} \end{equation}

According to formula (1), M×d sparrows form an initial population, and the concrete steps of initializing the sparrow population by using the cubic mapping chaos operator are as follows: To initialize the sparrow population, each sparrow undergoes iterative computations using the formula (2) to generate novel variable values within each dimension. Subsequently, these values are mapped to the sparrow population employing the function defined by formula (3). This procedure ensures the effective initialization of the sparrow population:

(3) \begin{equation} X_{i}=X_{lb}+\left(X_{lb}-X_{ub}\right)\times \left(y_{i}+1\right)\times 0\mathit{.}5 \end{equation}

In the context provided, $X_{lb}$ and $X_{ub}$ represent the lower and upper boundaries, respectively, of the search dimension for each sparrow individual. The variable $X_{i}$ denotes the resulting value of the individual variable after undergoing chaotic mapping. $y_{i}$ represents cubic sequences, a set of chaotic operators, and the cubic sequence can be mapped to sparrow individuals according to formula (3).

For the entrants, their foraging behavior mainly depends on the optimal position of the discoverer, and some entrants will always monitor the latter. Once the predation conditions are met, sparrows promptly displace themselves from their current positions to engage in competitive foraging for sustenance. The process of updating the location for an entrant is detailed as follows:

(4) \begin{equation} X_{i,j}^{t+1}=\left\{\begin{array}{l@{\quad}l} Q\cdot \exp \left(\dfrac{X_{\text{worst }}-X_{i,j}^{t}}{i^{2}}\right) & \text{ if }\quad i\gt n/2\\[12pt] X_{p}^{t+1}+\left| X_{i,j}^{t}-X_{P}^{t+1}\right| \cdot A^{+}\cdot L & \text{ otherwise } \end{array}\right. \end{equation}

Within the given context, the notation and variables are described as follows: $X_{p}^{t}$ represents the position of an individual with the best fitness value in the t-th iteration, while $X_{\text{worst}}$ corresponds to the position with the lowest global fitness value at the present moment. A denotes a 1×d matrix, and $A^{+}$ is the transpose of A multiplied by the inverse of the transpose, denoted as $A^{T}(A^{T})^{-1}$ . When i > n/2, it means that the participant has a low fitness value and has no right to fight for food. Consequently, they are obliged to explore new locations in search of foraging opportunities. Conversely, participants with an index i less than or equal to $n/2$ pursue foraging opportunities in proximity to the optimal position.

During the simulation experiment, a subset of sparrows, commonly known as “watchmen,” typically comprise approximately 10% to 20% of the total sparrow population. These watchmen fulfill the ongoing responsibility of diligently observing the sparrow population, promptly conveying early warning signals whenever perilous situations are detected. The primary objective behind these cautionary signals is to guide the entire population toward adopting anti-predation behaviors. The convergence rate of the standard vigilante position updating formula is reduced in the later period, and there is a local optimal constraint. Therefore, the number of vigilantes is dynamically adjusted by using an adaptive factor, and the search range is expanded by a larger number of vigilantes at the beginning of the algorithm. Toward the conclusion of the algorithm, reducing the number of watchmen can have advantageous effects, promoting algorithmic accuracy and facilitating enhanced local exploration capabilities. The dynamic decreasing formula of the number of watchmen is as follows:

(5) \begin{equation} \text{Num}=\mathrm{int}\!\left(\!\left(1-\frac{t}{t_{\max }}\right)\times \text{Num} _{ini}\right)+1 \end{equation}

where Num is the number of current watchmen, the symbol “t” represents the ongoing iteration count, “t _max” denotes the maximum allowable number of iterations, and “Num_ini” represents the initial count of watchmen.

The location update of the vigilante is described as follows:

(6) \begin{equation} X_{i,j}^{t+1}=\left\{\begin{array}{l@{\quad}l} X_{\text{best }}^{t}+\beta \cdot \left| X_{i,j}^{t}-X_{\text{best }}^{t}\right| & \text{ if }\quad f_{i}\gt f_{g}\\[9pt] X_{i,j}^{t}+K\cdot \left(\dfrac{\left| X_{i,j}^{t}-X_{\text{worst }}^{t}\right| }{\left(f_{i}-f_{w}\right)+\varepsilon }\right) & \text{ if }\quad f_{i}=f_{g} \end{array}\right. \end{equation}

Given the context, the notation and variables are defined as follows: $X_{\text{best}}$ represents the current global optimal position. β is employed as the step control parameter. K is a random number sampled from the interval [−1,1], and $f_{i}$ represents the fitness value of the current sparrow. Additionally, $f_{g}$ and $f_{w}$ are utilized to denote the best and worst fitness values, respectively. ε denotes a minimal constant value.

Upon reaching a stagnation point in the algorithm, a Gaussian random walk strategy is employed to generate fresh individuals. The generation formula for determining the position is as follows:

(7) \begin{equation} X_{i}^{t\mathit{+}1}=\mathrm{Gaussian}\!\left(X_{i}^{t},s\right) \end{equation}

(8) \begin{equation} \sigma = \cos\! \left(\pi \times \frac{t}{2t_{\max }}\right)\times \left(X_{i}^{t}-X_{r}^{\mathit{*}}(t)\right) \end{equation}

where $X_{r}^{\mathrm{*}}$ is a random individual in the discoverer population, and the step size is adjusted by a cosine function. As the number of iterations increases, there is a gradual reduction in disturbance, leading to an enhanced ability to adjust the optimization direction of the algorithm. Consequently, this results in an improvement in the search efficiency of sparrows.

The schematic representation of the enhanced SSA is illustrated in Fig. 4.

Figure 4. Flowchart of the algorithm.

3.2. Example experiment and result analysis

To validate the efficacy of the enhanced chaotic sparrow search algorithm (CSSA), comparative experiments were conducted, involving the ACO algorithm [Reference Liu, Hou, Tan, Liu and Song27], PSO algorithm [Reference Pozna, Precup, Horvath and Petriu28], and SSA algorithm [Reference Gao, Shen, Guan, Zheng and Zhang29]. To maintain fairness and consistency across the algorithms, a population size N = 30 and a maximum iteration count T = 500 were set. Table I provides a comprehensive overview of the specific algorithm parameters.

Table I. Algorithm parameters.

Table II provides a comprehensive overview of the essential details of each test function used.

Table II. Test function information table.

(1) Optimization and comparison of the unimodal test function.

F1 and F2 are unimodal test functions that have the ability to test the local development of the algorithm. To evaluate the performance of each algorithm, independent runs were conducted 50 times, with the average value and standard deviation serving as evaluation metrics. The optimization outcomes are presented in Table III, and the search space is shown in Fig. 5. It can be seen that, compared with the classical algorithm, CSSA is greatly improved, and the global optimal solution can be found for F1, and it performs better than other algorithms. The result obtained by CSSA for F2 is closer to the theoretical optimal solution, and the optimization result is more stable. Obviously, CSSA has stable performance and a better convergence effect, ranking first among these algorithms, and the improvement is of great significance. CSSA can enhance the quality and distribution uniformity of the initial population by introducing chaotic sequence mapping, which is helpful for the algorithm to conduct a more comprehensive search in the search space and achieve the purpose of improving the convergence accuracy and optimization performance of the algorithm. Chaos operators can usually better explore the search space and help to jump out of the local optimal solution by introducing randomness and nonlinear characteristics, which is very important for unimodal test functions. This balance of exploration and development ability is very important for finding the global optimal value in unimodal function.

Table III. The optimization results of the single-peak test function.

Figure 5. Search space of F1 and F2.

(2) Optimize the comparison on multimodal test function.

In addition to that, F3 and F4 represent multimodal test functions, characterized by a significant number of local extreme points. These test functions serve as suitable benchmarks for evaluating the algorithm’s optimization capability. Each algorithm was independently executed 50 times, and the optimization results are detailed in Table IV, while the convergence behavior is visually represented in Fig. 6. A noteworthy observation is that CSSA consistently converges toward the global optimal solution when dealing with F3, demonstrating superior stability compared to other algorithms. Notably, ACO exhibits the highest level of instability. Moreover, CSSA showcases improved stability and convergence speed in addressing the challenges posed by F4.

Table IV. The optimization results of the multi-peak test function.

Figure 6. Optimization curve of F3 and F4.

(3) Optimize the comparison on the fixed-dimension test function.

F5 and F6 are fixed-dimension test functions, which are suitable for comprehensive verification of global search and local exploration capability. Each algorithm has been independently run 50 times, and the optimization outcomes of the aforementioned algorithm are presented in Table V, while Fig. 7 depicts its convergence behavior. A close analysis of the data reveals that the convergence speed of CSSA on F5 is absolutely superior, and it effectively converges to the global optimal value, which is obviously superior to ACO, PSO, and SSA. On F6, CSSA can quickly converge to the optimal value.

Table V. The optimization results of the fixed-dimension test functions.

Figure 7. Optimization curve of F5 and F6.

In a single-peak test function, traditional sparrow algorithms may excessively rely on collective intelligence, leading to suboptimal performance in the global search process. In contrast, the improved sparrow algorithm introduces individual exploration behavior, enabling better local search and faster convergence to the extreme value of the unimodal function. Compared to ACO and PSO algorithms, the improved sparrow algorithm can adjust the level of cooperation among sparrows more flexibly, allowing the algorithm to better adapt to problem characteristics. In multimodal functions, there are multiple local optima. The improved sparrow algorithm has the ability to escape local optima to some extent. This is because the improved sparrow algorithm incorporates the cubic mapping chaotic operator into the basic sparrow algorithm, facilitating better initialization of the sparrow population. This ensures that sparrow individuals traverse all positions, thereby enhancing the robustness and optimization performance of the algorithm.

To sum up, the improved sparrow algorithm performs better in both unimodal and multimodal functions, primarily due to its diverse search strategies and a balance between collective intelligence and individual exploration. To sum up, CSSA has good optimization performance and obvious competitive advantage regardless of unimodal, multimodal, and fixed-dimension test functions.

4.1. Construction of truss space model

The spatial truss is composed of truss rods and truss joints. We take the geometric centers of 96 truss joints as path nodes and input the coordinate data of these nodes into Matlab to generate the mathematical model of the spatial truss and establish the three-dimensional spatial coordinate system of the model, as shown in Fig. 8. The coordinates of some truss joints are shown in Table VI. The starting point of setting path optimization is truss joint No.1, and the target point is truss joint No.96. Then the target task is to find the shortest path from node 1 to node 96 in the three-dimensional space model.

Figure 8. Truss mathematical model.

Table VI. Coordinates of some truss joints.

4.2. Path planning algorithm flow

In this section, CSSA is used to solve the three-dimensional path planning problem and find out the shortest path of the climbing movement of the in-orbit cellular robot in a truss environment. The specific algorithm flow steps based on CSSA are as follows:

Step 1: Import the coordinate information of each node and establish the coordinate model of the space truss.

Step 2: Initialize parameters and set the maximum number of iterations of the algorithm, safety threshold, etc.

Step 3: Set the number of population discoverers, watchmen, and participants.

Step 4: Set the starting point and target point.

Step 5: Update the positions of discoverer, vigilante, and entrant to get the target path.

Step 6: Conduct a comparison between the current path and the previous path. In the event that the current path proves to be shorter, update the optimal path accordingly.

Step 7: Evaluate whether the termination condition has been met. If not, proceed to repeat steps 5 and 6. Conversely, if the termination condition is satisfied, the optimal path is to be presented as the final output.

4.3. Path planning simulation and analysis

In this section, Matlab is used to simulate the path planning of an on-orbit cellular robot. To evaluate the performance of the enhanced SSA, we utilized the ACO algorithm, the PSO algorithm, and the conventional SSA for on-orbit cellular robot path planning.

As shown in Table VII, for CSSA and SSA, the population size was set to n = 50, with a maximum of 50 iterations. The safety threshold (ST) was determined as 0.8, and the discoverer accounted for 20% of the population size, while the number of sparrows aware of danger was set to 10. ACO utilized parameters “α = 1” and “β = 5,” with a population size of Nc = 50 and a maximum iteration count of 50. As for PSO, the inertia weight was assigned as 0.729, while the social learning factor (c1) and individual learning factor (c2) were both set to 1.49445. The starting point of the path planning was designated as truss joint 1, with truss joint 96 serving as the target point. In order to enhance path planning efficiency, this simulation study disregarded the actual volume of the robot, resulting in simulation-obtained paths that align with the truss model.

Table VII. Initial parameter settings.

The optimization outcomes of the ACO path planning are presented in Fig. 9. The simulation results demonstrate that the algorithm yielded the shortest path nodes as follows: “1 → 49 → 55 → 56 → 62 → 63 → 69 → 70 → 76 → 77 → 83 → 84 → 90 → 96,” with an optimal path length of 12,628.74 mm. Figure 10 exhibits the optimization results of PSO path planning, illustrating that the shortest path nodes acquired by this algorithm are “1 → 8 → 15 → 22 → 29 → 36 → 42 → 48 → 96,” with an optimal path length of 9719.42 mm. Moreover, the results of the SSA path planning optimization (Fig. 11) indicate that the shortest path nodes obtained by this algorithm align with the previous PSO path, namely “1 → 8 → 15 → 22 → 29 → 36 → 42 → 48 → 96,” and the optimal path length remains at 9719.42 mm. Finally, Fig. 12 showcases the optimization outcomes of CSSA path planning, revealing that the shortest path nodes derived from this algorithm coincide with the path obtained by PSO and SSA, specifically “1 → 8 → 15 → 22 → 29 → 36 → 42 → 48 → 96.” The optimal path length remains unchanged at 9719.42 mm.

Figure 9. ACO path planning finder results.

Figure 10. PSO path planning finder results.

Figure 11. SSA path planning finder results.

Figure 12. CSSA path planning finder results.

Through comparative analysis, it is evident that PSO, SSA, and CSSA all demonstrate the capability to identify an optimal path. However, ACO tends to find an optimal path that is characterized by a prolonged distance. This discrepancy arises due to the ACO algorithm’s susceptibility to getting trapped in local optimal solutions during the early stages, impeding its ability to escape. The convergence behavior of each algorithm’s robot path planning is depicted in Fig. 13. Notably, from Fig. 13, it is apparent that CSSA exhibits a faster convergence rate compared to PSO and SSA. Moreover, the curve exhibits smooth variations with minimal amplitude.

Figure 13. Iterative curves of algorithms.

To ascertain the effectiveness and superiority of CSSA, the iteration times and population number are selected as independent variables, and the standard SSA is set as the experimental control group. The optimal path and reaction time are comprehensively evaluated, and the running results are shown in Table VIII.

Table VIII. Algorithm comparison.

Based on the findings in Table VIII, it is evident that the optimal path index exhibits significant improvement as the number of iterations and population increases. Specifically, when the number of iterations and population reaches approximately 20, CSSA successfully generates the shortest path for the in-orbit cellular robot traversing the truss structure. At this point, the shortest path length measures 9719.42 mm, with an average path length of the same value. The corresponding reaction time is 7.31 s, surpassing SSA by 0.05% in this regard while achieving an 11.18% reduction in the optimal reflection time. Conversely, SSA only converges to an optimal path when the number of iterations and population is approximately 30.

Analyzing the standard deviation index of the optimal path reveals a notable issue with the standard SSA algorithm when confronted with path optimization challenges. Specifically, the algorithm is prone to local optimal solutions, resulting in substantial deviation of the optimal path data for the in-orbit cellular robot on the truss when the iteration number and population are set at lower values, which leads to a large deviation of the optimal path data of the in-orbit cellular robot on the truss when the iteration number and population number are low, and CSSA effectively avoids this problem. According to the reaction time index, it can be seen that the introduction of CSSA-improved strategies increases part of the running time, but it does not introduce additional time error to the overall performance of path optimization and still maintains high optimization stability.

In order to show the performance gap between CSSA and SSA in more detail, the optimization results are shown in Table IX.

Table IX. Comparison of SSA and CSSA indicators.

From Table IX, it can be observed that the optimal path obtained by CSSA for in-orbit cellular robots remains consistent with the standard SSA. The overall reaction time has increased by 51.60%, and the effective iteration count has increased by approximately 13.87% compared to the standard SSA.

In conclusion, the enhanced SSA has achieved improvements in convergence speed and optimization stability to a certain extent, significantly reducing the number of iterations required to obtain the optimal path for in-orbit cellular robots compared to the standard SSA. The introduction of enhanced algorithmic strategies has had no additional impact on time complexity and space complexity, maintaining their consistency. Consequently, it can be inferred that the enhanced strategies effectively enhance the search performance and optimization efficiency of the algorithm, making the ISSA well suited for addressing the path search problem in on-orbit truss cellular robots.