Design and Mechanical Behavior Research of Highway Guardrail Patrol Robot

Abstract

Conducting risk assessments on highways is a critical task. This paper introduces a mobile platform designed for guardrail inspection robots to address the gap between the inspection requirements in road traffic management and the current capabilities of existing highway inspection robots. The platform is utilized for random vehicle inspections, road environment assessments, and transportation equipment evaluations. The robot is designed to operate on double-waveform beam guardrails and features an innovative adaptive dual-wheel tensioning mechanism, significantly enhancing its ability to adapt to the guardrail’s shape and joints. A mechanical model of the robot was developed, and the impact of the tension on the robot’s obstacle-crossing performance was analyzed and optimized through theoretical and simulation-based studies. Finally, a prototype of the robot was constructed, and a testing platform for the highway guardrails was established to evaluate the robot’s operational capabilities. The results demonstrate that the robot exhibits excellent performance in both operation and obstacle-crossing tasks.

1. Introduction

With the rapid development of road traffic systems, traffic accidents have become an inevitable concern. Manual inspections, while commonly used, are often time-consuming, labor-intensive, and inefficient, and pose significant safety risks. Although fixed road testing methods reduce some of these risks, their coverage is limited and often insufficient for comprehensive assessments. In recent years, numerous studies have explored the application of robotics in the field of road traffic. For instance, in the area of bridge detection [1,2,3,4,5,6,7], robots equipped with cameras have been developed to scan and inspect entire bridge structures. Prateek Prasanna et al. pioneered the application of automatic crack detection in robotic bridge scanning, proposing a method for the automatic identification of concrete bridge cracks. This approach enables robots to analyze data and quickly assess bridge safety after completing the scans [8]. Duan et al. addressed the issue of limited measuring points in bridge safety monitoring by introducing a global displacement detection method based on natural texture features. This method fundamentally resolves the problem of damage identification caused by sparse monitoring data [9]. Additionally, Divija Swetha Gadiraju et al. proposed a novel approach using Deep Reinforcement Learning (DRL) to control unmanned aerial vehicles (UAVs) for bridge crack detection. By integrating Canny Edge detection and Convolutional Neural Networks (CNNs), their method demonstrates superior detection quality compared to traditional techniques [10].

Robots have also been widely applied in the field of tunnel detection [11,12,13,14,15]. Elisabeth et al. developed a robotic system called ROBO-SPECT for highway tunnel inspections. This system enables the accurate, rapid, and reliable inspection and evaluation of tunnel linings under safer working conditions and during continuous traffic operations [16]. Similarly, Wang et al. designed a suspension robot equipped with multiple sensors for autonomous tunnel detection and monitoring. This robot is capable of collaborating with firefighters and rescue personnel to perform timely rescue missions in emergency situations [17].

Robots are also being utilized in road surface detection [18,19,20,21,22]. Lu et al. proposed a method based on layered visual sensors to achieve robust detection in complex road environments [23]. Fukuhara et al. introduced a flood-filled road detection method capable of adapting to various road types. This method allows robots to select the appropriate controllers based on the road surface conditions or safe routes [24]. Jing et al. reviewed the advancements in railway inspection robots and conducted a feasibility analysis by testing a typical track inspection robot prototype on a specific railway line [25]. Additionally, Cheng et al. developed a curb detection method for mobile robots, enabling them to distinguish between roads and sidewalks in urban residential areas using stereo vision technology [26].

Although extensive research has been conducted on the application of robots in the field of road traffic, there remains a significant gap in the development of robots specifically designed for highway patrol, inspection, and surveillance. Currently, the most widely deployable robotic technology for such tasks is the unmanned aerial vehicle (UAV). However, UAVs face critical limitations, including short endurance times and restricted payload capacities [27,28]. To effectively serve the road traffic sector, it is essential to fundamentally improve the endurance and load-bearing capabilities of UAVs.

Ground-based robots generally offer significantly longer endurance compared to aerial systems, and can be equipped with charging stations for autonomous recharging. This capability allows them to operate stably even under harsh weather conditions, such as rain, snow, and sandstorms. In the context of highways, researchers are actively exploring the use of ground robots to replace human labor for routine patrol, inspection, and surveillance tasks. While some patents and related reports have been published, there has been no comprehensive analysis of the robots’ performance. This lack of detailed evaluation suggests that the technology in this field remains relatively underdeveloped.

This paper addresses the gap between the operational requirements of road traffic surveillance and the current capabilities of highway patrol robots. Through a comprehensive analysis of the strengths and limitations of various inspection, patrol, and surveillance robots, and by considering the unique conditions of highway environments, a mobile platform for a guardrail inspection robot specifically tailored for highway applications is proposed. Furthermore, the operational performance of the robot’s mobile platform is thoroughly analyzed and validated. This ensures that the robot can operate on highways for extended periods in a stable, efficient, safe, and reliable manner.

2. Mechanism System Description

2.1. Structural Design

When the robot is patrolling on the guardrail, its passability and stability must be ensured while preventing it from detaching from the guardrail. Based on the shape and dimensions of the double-corrugated beam guardrail, this study designs a mobility mechanism tailored to fit the guardrail. The mechanism incorporates an adaptive bilateral-wheel tensioning structure, ensuring a secure and stable fit on the guardrail. The driving system employs a simpler unilateral driving mode, which not only reduces the robot’s weight, but also simplifies its overall structure. As shown in Figure 1a, the guardrail inspection robot primarily consists of the following components: the main frame, driven wheel system, non-driven wheel system, guide shaft system, auxiliary wheels, load wheels, and shock-absorbing tensioning springs.

The main frame serves as the structural foundation, supporting all other components of the robot. To ensure lightweight construction and ease of assembly, the frame is constructed using aluminum profiles. Given that the guardrail surface is relatively flat and smooth, the robot employs a wheeled mobility system. The driving mechanism utilizes a single-sided driving configuration. To enhance the transmission efficiency and protect the guardrail, the driven wheel is designed as a rubber-coated wheel that conforms to the guardrail’s shape. This wheel is paired with a unilateral auxiliary wheel on the opposite side of the guardrail, forming a bilateral structure tightened by a spring, as illustrated in Figure 1b. The non-driven wheel system mirrors the structure of the driving gear train but does not include a drive mechanism, serving instead to balance the robot. To prevent the robot from detaching from the guardrail, a set of load wheels is incorporated to assist the shock-absorbing tension springs in counteracting gravitational forces.

2.2. Analysis of Falling Conditions and Guardrail Safety Conditions

Since the guardrail inspection robot employs a tight-fitting double-sided wheel structure, a spring mechanism is required to provide the appropriate tension. This ensures effective power transmission while maintaining the robot’s mobility and operational stability. The single-sided drive design also results in differing pulling force requirements on the driven and non-driven wheel sides. By adjusting the stiffness k₁ of the spring on the driving wheel side and the stiffness k₂ of the spring on the driven wheel side, the forces exerted on the wheel frames on both sides can be fine-tuned. The gravity distribution coefficients for the master and slave sides are defined as λ, which satisfies the following relationship:

$$\lambda ={\displaystyle \frac{k_1}{k_1+k_2}},$$

(1)

When the robot is in a critical situation where it is at risk of detaching from the guardrail, it is assumed that the load wheels no longer counteract the gravitational force. In this scenario, the entire weight of the robot is supported by the rubber-coated wheels on both the master and slave sides, as illustrated in Figure 1c. At this point, the following equilibrium equation applies:

$$\left\{\begin{array}{l}\left({F_N}^{\prime }+{F_N}^{″}\right)\cos \beta +\mu {F_N}^{″}\sin \beta -F^{\prime }=0\\ \left({F_N}^{\prime }-{F_N}^{″}\right)\sin \beta +\mu {F_N}^{″}\cos \beta +\lambda G=0\end{array}\right.,$$

(2)

When the wheel is in a critical falling-off state, where F_N′ = 0, the spring tension on the driven wheel frame should satisfy the following formula:

$$F^{\prime }+F^{″}>G({\displaystyle \frac{\cos \beta }{\sin \beta }}+\mu ),$$

(3)

Here, G represents the self-weight of the robot. The target mass for the patrol robot developed in this study is set at 30 kg. μ denotes the static friction coefficient between the rubber-coated wheel and the corrugated beam guardrail, and β is the angle between the hypotenuse of the rubber-coated wheel and the vertical direction. F′ and F″ represent the forces required for the driven wheel frame and the non-driven wheel frame, respectively, to prevent slippage. These forces depend on the spring stiffness values k₁ and k₂, the horizontal distance x, the vertical distance h₀, and the inclination angle γ of the spring. The relationship is expressed as follows:

$$\left\{\begin{array}{l}{F}^{\prime }=4k_1(x-x_0\cos \gamma )\\ F^{″}=4k_2(x-x_0\cos \gamma )\\ h_0/x=\tan \gamma \end{array}\right.,$$

(4)

x₀ in the formula is the initial length of the spring. In order to control the variables, the impact of the spring stiffness on the robot performance is analyzed separately, so the initial length of the spring on the active side and the driven side is ensured to be consistent. By changing the spring stiffness, the spring-tensioning force of the driving gear train and the driven gear train is changed. Based on the dimensions of the corrugated beam and the compatibility of the mechanical structure, the preset values for each parameter are shown in

Table 1. Structural parameter preset values.

Parameter	Value
Angle between edge of rubber wheel and vertical direction β	60°
Horizontal distance between both ends of spring x	185 mm
Vertical height at both ends of spring h0	62.5 mm
Initial length of spring x0	100 mm
Frictional coefficient μ	0.8 N/mm

.

Based on the above analysis, it can be concluded that, to ensure the inspection robot does not detach from the guardrail while operating on a stable track, the following relationship must be satisfied:

$$k_1+k_2>1.21,$$

(5)

If the sum cannot satisfy the aforementioned inequality, the robot will detach from the guardrail. It is important to note that the value of 1.21 is calculated based on the initial structural parameters of the robot and is specific to this configuration. Different robot dimensions may result in different corresponding values.

Given that the auxiliary wheels must continuously traverse the connecting joints on the back of the guardrail during operation, using rubber-coated wheels would result in significant wear. This would not only reduce the robot’s operational time on the guardrail, but also negatively impact its overall performance. To address this, the single-sided wheels are constructed from 45# steel and serve as the auxiliary wheels for the robot. However, this introduces a potential risk to the guardrail’s safety, necessitating the use of tension springs to provide a pulling force within a safe range. This condition can be expressed as:

$$\sigma ={\displaystyle \frac{F_N}{A}}\le \left[\sigma \right],$$

(6)

Corrugated beam guardrails are typically constructed from Q235 steel, which has an allowable stress [σ] of approximately 160 MPa. The contact area A between the auxiliary wheels and the guardrail is about 3 mm². It can be obtained as follows:

$$k_i\le 2.4(i=1,2),$$

(7)

By analyzing the conditions required to prevent the guardrail inspection robot from detaching and the safety conditions of the guardrail, the boundary conditions for the spring stiffness k₁ on the driven wheel side and the spring stiffness k₂ on the non-driven wheel side were determined. Within these boundary conditions, k₁ and k₂ are allocated appropriately, and the robot’s operational performance is further analyzed.

3. Analysis of Robot Obstacle-Crossing Behavior

3.1. Description of Guardrail Obstacles

The guardrail inspection robot proposed in this study operates on a double-waveform beam guardrail, assuming the guardrail is in good working condition. Guardrails with significant deformations caused by collisions, corrosion, or fractures due to natural factors are not considered in this analysis. To ensure the robot can smoothly complete its tasks while minimizing the damage to the guardrail during operation, its obstacle-crossing performance is analyzed under the condition of no initial speed.

Many factors influence the actual working conditions of the robot’s obstacle-crossing performance. To simplify the analysis, the following assumptions are made:

(1) The weight of the non-structural components, such as motors and controllers, is neglected, and the overall mass of the robot is simplified as a concentrated mass. Assuming the robot is symmetrical along its vertical axis, the three-dimensional spatial problem is reduced to a two-dimensional planar problem;

(2) The robot’s frame, obstacle-crossing wheels, auxiliary wheels, and other components are treated as rigid bodies, with deformation factors disregarded in the analysis.

(3) The damping factor of the spring is neglected, and it is assumed that the support reaction force acting on each wheel is linearly proportional to the spring deformation, with the proportionality constant k.

Based on the above assumptions, the obstacle crossing of the robots can be divided into the following four situations: front wheel obstacle crossing of auxiliary wheels (FAOC), rear wheel obstacle crossing of auxiliary wheels (RAOC), obstacle crossing of non-driven rubber-coated wheels (NDOC), and obstacle crossing of driven rubber-coated wheels (DOC). As shown in Figure 2, the red wheel represents the driving wheel, the arrow indicates the direction of travel, and the yellow fill represents the wheel that is crossing obstacles.

3.2. Analysis of Obstacle-Crossing Behavior

When the robot is in the front auxiliary wheel obstacle-crossing state (FAOC), the force analysis is illustrated in Figure 3a. The static equilibrium equations are as follows:

$$\left\{\begin{array}{l}\phi F_N-{\mu }_1{F}_{N2}-\mu F_{N3}-F_{N1}\cos \alpha -{\mu }_1{F}_{N1}\sin \alpha =0\\ F_N+F_{N3}-F_{N2}+{\mu }_1{F}_{N1}\cos \alpha -F_{N1}\sin \alpha =0\\ {\mu }_1{F}_{N1}r+F_{N2}l+{\mu }_1{F}_{N2}r-F_{N3}l+\mu F_{N3}\left(a+r\right)-\phi F_N\left(a+r\right)=0\end{array}\right.,$$

(8)

By setting μ, μ₁ = 0, the above equations can be simplified as follows:

$${\displaystyle \frac{h}{r}}=1-{\displaystyle \frac{1}{\sqrt{1+{\left(\frac{\phi l}{\phi \left(a+r\right)-l}\right)}^2}}},$$

(9)

In this formula, r is the radius of the auxiliary wheel, μ₁ is the sliding friction coefficient between the auxiliary wheel and the guardrail, and R is the theoretical contact radius of the contact circle tangent to the theoretical contact line between the rubber-coated wheel and the guardrail; μ is the sliding friction coefficient between the non-driven rubber wheel and the guardrail, φ is the adhesion coefficient between the rubber wheel and the guardrail, a is the distance between the rear end of the corrugated beam guardrail and the theoretical contact line, and l is the track width.

The mechanical analysis of the driven rubber-coated wheel obstacle-crossing scenario (DOC) is illustrated in Figure 3a, and the static equilibrium equations are as follows:

$$\left\{\begin{array}{l}\phi F_N\sin \alpha -F_N\cos \alpha -\mu F_{N3}-{\mu }_1{F}_{N1}-{\mu }_1{F}_{N2}=0\\ \phi F_N\cos \alpha +F_N\sin \alpha -F_{N1}+F_{N3}-F_{N2}=0\\ \phi F_{N}r+F_{N3}l+\mu F_{N3}\left(R-h\right)-{\mu }_1{F}_{N1}(a+r)-{\mu }_1{F}_{N2}(a+r)-F_{N2}l=0\end{array}\right.,$$

(10)

By setting μ, μ₁ = 0, the above equations can be simplified as follows:

$${\displaystyle \frac{h}{R}}=1-{\displaystyle \frac{1}{\sqrt{1+{\phi }^2}}},$$

(11)

Similar to the obstacle-crossing analysis of the front wheel of the auxiliary wheel, the mechanical analysis of the rear wheel of the auxiliary wheel in the case of obstacle crossing (RAOC) is shown in Figure 3b.

The static analysis equations are as follows:

$$\left\{\begin{array}{l}\phi F_N-{\mu }_1{F}_{N1}-\mu F_{N3}-{\mu }_1{F}_{N2}\sin \alpha -F_{N2}\cos \alpha =0\\ {\mu }_1{F}_{N2}\cos \alpha -F_{N2}\sin \alpha +F_N+F_{N3}-F_{N1}=0\\ {\mu }_1{F}_{N2}r+{\mu }_1{F}_{N1}\left(r-h\right)+\mu F_{N3}(a+r)-\phi F_N(a+r)\\ +F_{N3}(a+R+r)\tan \theta -F_{N1}l\cos \theta \\ +F_N\left[l\cos \theta +(a+R+r+h)\tan \theta \right]=0\end{array}\right.,$$

(12)

By setting μ, μ₁ = 0, the above equations can be simplified as follows:

$$F_N[-\phi (a+r)+\phi l\tan \alpha \cos \theta +(a+R+r+h)\tan \theta ]+F_{N3}[(a+R+r)\tan \theta -l\cos \theta ]=0,$$

(13)

Due to the deflection of the wheel frame during the obstacle crossing, the equation includes a trigonometric function that cannot be simplified, making it too complex to solve numerically. Drawing on prior research experience related to the obstacle-crossing performance of automobiles, this study transforms the above model. The static equilibrium equations for the transformed model are as follows:

$$\left\{\begin{array}{l}\phi F_N-{\mu }_1{F}_{N1}-\mu F_{N3}-{\mu }_1{F}_{N2}\sin \alpha -F_{N2}\cos \alpha =0\\ {\mu }_1{F}_{N2}\cos \alpha -F_{N2}\sin \alpha +F_N+F_{N3}-F_{N1}=0\\ {\mu }_1{F}_{N2}r+{\mu }_1{F}_{N1}r+\mu F_{N3}(a+r)-\phi F_N(a+r)-F_{N1}l+F_{N}l=0\end{array}\right.,$$

(14)

By setting μ, μ₁ = 0, the above equations can be simplified as follows:

$$F_N[\phi (a+r)/l-\phi \tan \alpha ]+F_{N3}=0,$$

(15)

At this time, the supporting forces F_N and F_N3 perpendicular to the direction of the guardrail are applied to the driven and non-driven rubber-coated wheels. Through analysis, it can be concluded that

$$\left\{\begin{array}{l}{F}_N=F_k=k_1(x-x_0)\\ F_{N3}=F_{k3}=k_2(x-x_0)\end{array}\right.,$$

(16)

Substituting into the equations yields the following:

$${\displaystyle \frac{h}{r}}=1-{\displaystyle \frac{1}{\sqrt{1+{\left(\frac{k_1\phi l}{k_1\phi \left(a+r\right)+k_{2}l}\right)}^2}}},$$

(17)

The mechanical analysis of the non-driven rubber-coated wheel obstacle-crossing scenario (NDOC) is illustrated in Figure 3c. The static equilibrium equations are as follows:

$$\left\{\begin{array}{l}\phi F_N-{\mu }_1{F}_{N2}-{\mu }_1{F}_{N1}-F_{N3}\cos \alpha -\mu F_{N3}\sin \alpha =0\\ F_N-F_{N2}-F_{N1}-\mu F_{N3}\cos \alpha +F_{N3}\sin \alpha =0\\ \mu F_{N3}r+F_{N}l-\phi F_{N}r-F_{N1}l+{\mu }_1{F}_{N2}\left(a+r\right)+{\mu }_1{F}_{N1}\left(a+r\right)=0\end{array}\right.,$$

(18)

By setting μ, μ₁ = 0, the above equations can be simplified as follows:

$$\phi F_N=F_{N3}\cos \alpha ,$$

(19)

F_N and F_N3, respectively, represent the support forces perpendicular to the guardrail direction on the driven and non-driven rubber-coated wheels. The magnitude of the force is shown in Figure 3c, where F_k, F_k3, F_N, and F_N3 represent the spring tension and support force of the guide shaft on the active and driven wheel systems, respectively. Through analysis, it can be concluded that

$$\left\{\begin{array}{l}{F}_N=F_k=k_1(x-x_0)\\ F_{N3}={\displaystyle \frac{F_{k3}}{\sin \alpha }}={\displaystyle \frac{k_2(x-x_0)}{\sin \alpha }}\end{array}\right.,$$

(20)

Substituting into the equations yields the following:

$${\displaystyle \frac{h}{R}}=1-{\displaystyle \frac{1}{\sqrt{1+{(\frac{k_1}{k_2})}^2{\phi }^2}}},$$

(21)

Based on the above analysis, it can be concluded that the obstacle-crossing performance and adhesion coefficient of the robot φ, the distance a from the rear end of the corrugated beam guardrail to the theoretical contact line, the wheelbase l, the radius r of the auxiliary wheel, the theoretical contact circle radius R of the contact line between the rubber-coated wheel and the guardrail, and the spring stiffness k₁ on the active wheel side and k₂ on the passive wheel side are related, as shown in

Table 2. Impact parameters of obstacle crossing.

	a	l	r	R	φ	k1	k2
FAOC	√	√	√	—	√	—	—
RAOC	√	√	√	√	√	√	√
NDOC	—	—	—	√	√	√	√
DOC	—	—	—	√	√	—	—

(“√” indicates that the parameter is related to obstacle-crossing performance, while “—” indicates that it is unrelated).

3.3. The Influence of Different Parameters on Obstacle-Crossing Behavior

The obstacle-crossing height is influenced by numerous parameters, making it challenging to study and analyze effectively. Parameters such as the distance a from the rear end of the corrugated beam guardrail to the theoretical contact line, the wheelbase l, the radius r of the auxiliary wheel, and the theoretical contact circle radius R of the contact line between the rubber-coated wheel and the guardrail are affected by the structure. Research on the impact of these parameters on the obstacle-crossing performance is relatively well established and is not the primary focus of this study. Therefore, in line with the requirements for compactness and coordination of the mechanical structure, fixed values are assigned to the parameters not under investigation, as shown in

Table 3. Fixed-value parameter values.

Parameter	Value
Distance between edge of guardrail and theoretical contact line a	80 mm
Auxiliary wheel radius r	30 mm
Theoretical contact circle radius for rubber-coated wheels R	55 mm
Wheelbase l	180 mm

. Although the influence of the adhesion coefficient φ on the obstacle-crossing performance has also been extensively studied, this article includes an analysis of φ due to its strong independence from the specific structural configurations.

Regarding the coefficient of adhesion φ, the spring stiffness k₁ on the active wheel side, and the spring stiffness k₂ on the passive wheel side, the control variable method is employed to analyze the influence of these parameters on the obstacle-crossing height. When analyzing the influence of a certain parameter on the obstacle-crossing performance, the other parameters are held constant at fixed values (φ = 0.8, k₁ = 2, k₂ = 0.5).

The results of the theoretical analysis, as shown in Figure 4a, indicate that as the adhesion coefficient φ increases, the obstacle-crossing height h for the front and rear auxiliary wheels, non-driven rubber-coated wheels, and driven rubber-coated wheels improves. Among these, the driven rubber-coated wheel exhibits the lowest obstacle-crossing height. This is due to the robot’s single-sided driving mechanism: when the sole driving wheel crosses an obstacle, the spring consumes a significant amount of energy, resulting in the lowest obstacle-crossing height. When the adhesion coefficient φ reaches 0.8, the obstacle-crossing height of the driven rubber wheel can reach approximately 10 mm, satisfying the obstacle-crossing requirements (Through investigation, we have preset 10 mm as the reference value for the height of the guardrail obstacle, which is indicated in the green area of Figure 4).

As shown in Figure 4b, it is evident that as the stiffness k₁ of the active wheel side spring increases, the obstacle-crossing height of the rear auxiliary wheel and the non-driven rubber-coated wheel also increases. Conversely, when the stiffness k₂ of the passive wheel side spring increases, the obstacle-crossing performance of the rear auxiliary wheel and the non-driven rubber-coated wheel gradually decreases. This occurs because the non-driven rubber-coated wheel does not contribute to the robot’s propulsion. Increasing the stiffness of the driven wheel side spring introduces additional resistance to the robot’s movement.

From the above analysis, it can be concluded that to ensure an obstacle-crossing height greater than 10 mm, the stiffness k₂ of the non-driven wheel side spring should be minimized, while the stiffness k₁ of the driven wheel side spring should be maximized. However, these selections must also comply with the detachment conditions and the safety requirements of the guardrail. Therefore, while ensuring that the robot neither detaches from the guardrail, nor fails to meet the obstacle-crossing requirements during operation, springs with lower stiffness should be used whenever possible. Based on this analysis, five sets of parameter combinations were determined, and their specific values are provided in

Table 4. Parameters of each group of spring stiffness.

(Unit N/mm)	k1	k2
Group 1	2	0.1
Group 2	2	0.3
Group 3	2	0.5
Group 4	2.2	0.5
Group 5	2.4	0.5

.

4. Dynamic Simulation of Guardrail Inspection Robot

4.1. Barrier-Free Guardrail Section

The movement of the robot along the barrier-free section is relatively straightforward, consisting of two parts: movement on a straight section and movement on a small-curvature curved section. For the small-curvature curved sections, guardrails are typically treated using compression bending, allowing the two sections to be combined. The robot is set to travel back and forth along the combined guardrail section once, with a one-way travel time of 17 s and a total duration of 34 s.

The simulation results, as illustrated in Figure 5, show the vertical position changes of the robot during its operation in the barrier-free section. It is evident that the robot’s vertical position is not stable or constant, but fluctuates continuously during operation. This occurs because the spring constraint is not entirely rigid. Additionally, the curve is largely symmetrical, indicating minimal differences in stability between the robot’s forward and backward (return) movements along the same guardrail section.

When the stiffness k₁ of the active wheel side spring and the stiffness k₂ of the passive wheel side spring are set to the values of Group 2, Group 3, and Group 4, the robot runs the most smoothly. On the small-curvature bending guardrail section, the robot experiences a downward movement due to the compression of the load-bearing wheel and the bending guardrail, but it automatically adjusts after returning. When set to Group 1, the robot experienced a slight drop at 12 s, but overall, the stability was within an acceptable range. When k₁ and k₂ are set to 2.4 and 0.5, the robot experiences a significant drop at 7–8 s, and the stop of the fall is due to the limiting effect of the single edge of the auxiliary wheel. This occurs because the robot is driven on one side, and the excessively high spring stiffness on the active wheel side exerts a stronger force on the guardrail than the overall structure on the passive wheel side. As a result, the robot cannot automatically adjust to a neutral position and instead becomes supported by the active wheel side, causing displacement.

In the theoretical analysis, to simplify the mathematical model, the friction coefficient μ₁ between the auxiliary wheels and the guardrail, as well as the friction coefficient μ between the non-driving rubber-coated wheels and the guardrail, are neglected. Additionally, when analyzing the detachment conditions of the robot on the guardrail, the robot’s mass distribution is idealized as a planar problem, with the center of gravity approximated at the centroid of the active and driven axes. While the simulation environment is closer to real-world conditions, setting a higher stiffness k₁ for the active wheel side spring, though theoretically beneficial for the obstacle crossing, can lead to excessive force on the active wheel side in a three-dimensional environment. This makes it difficult for the robot to adapt to guardrail variations, increasing the risk of derailment. Consequently, the analysis of the spring stiffness setting for Group 5 is excluded in the following discussion.

4.2. Barrier Guardrail Section

The obstacles of guardrail groups are theoretically divided into three types: straight obstacle section, slope obstacle section, and turning obstacle section. A straight obstacle refers to the distance d generated when two sections of the guardrail are spliced together due to shape limitations, generally ranging from 5 to 10 mm in size. The simulation uses four sections of the guardrail spliced, with d values of 7.5 mm, 10 mm, and 12.5 mm, as shown in Figure 6a.

When the stiffness k₁ of the active wheel side spring and the stiffness k₂ of the driven wheel side spring are set according to the parameters of Groups 1–4, the displacement curve of the robot in the X-direction is shown in Figure 6b. It can be observed that when the spring parameters are set to Group 1, the robot experiences jamming during the return journey. This occurs because the stiffness of the spring on the driven wheel side is too weak to provide sufficient force to maintain the robot’s alignment with the guardrail. Over time, this misalignment accumulates, and once it reaches a critical level, the excessive angle between the robot and the guardrail causes jamming. As a result, the spring parameter settings of Group 1 are considered unreasonable and are excluded from further consideration. When the spring parameters are set to Groups 2–4, the robot successfully completes back-and-forth movement on the straight obstacle guardrail group, performing four obstacle-crossing behaviors: front auxiliary wheel obstacle crossing, rear auxiliary wheel obstacle crossing, non-driven rubber-coated wheel obstacle crossing, and driven rubber-coated wheel obstacle crossing.

By observing the vertical (Y-direction) displacement curve of the robot under the parameter settings of Groups 2–3, as shown in Figure 6c, it is evident that the vertical displacement values are relatively small and exhibit no abrupt changes when the robot is not crossing obstacles. This demonstrates that the robot can maintain overall stability during the obstacle crossing. However, under different spring parameters, the vertical displacement of the robot varies when the rubber-coated wheel crosses obstacles, with the fourth group of parameters showing the least variation. This occurs because the force on the active side of the robot is relatively large, restricting its movement. Unlike in the obstacle-free section, a larger vertical displacement during the obstacle crossing is beneficial, as it reduces the robot’s compression on the guardrail and better protects the guardrail’s structural integrity.

To comprehensively evaluate the robot’s performance, it is also necessary to consider two additional types of obstacles: slope obstacles and turning obstacles. Turning obstacles occur on high-curvature curved sections and arise due to angular deviations of the guardrail in the vertical direction. It is reasonable to conclude that such deviations only affect the obstacle-crossing height and are essentially equivalent to straight obstacles. Therefore, they will not be analyzed separately.

When the road environment of the expressway includes slopes, the angles between the guardrail groups must be adjusted. By accumulating multiple small angles, the guardrail groups can be aligned with the slope of the roadside. The slope obstacle is set at an angle of 1°, which can be categorized into two scenarios: uphill and downhill, as illustrated in Figure 7a.

When k₁ is set to 2 and k₂ is set to 0.5, the robot can smoothly cross both the uphill and downhill slope obstacles. Its displacement in the X- and Y-directions is shown in Figure 7b,c, respectively. The other two parameter groups exhibit jamming phenomena, and the reason is the same as previously described: a mismatch in forces between the active and passive sides. Through analysis, it can be observed that the displacement curve on the slope obstacle guardrail section is symmetrical, further proving that the robot exhibits excellent adaptability to the guardrail under the third group of spring stiffness parameters.

5. Experiment

5.1. Prototype Development

The overall framework of the guardrail inspection robot developed in this study is constructed using 20 mm × 20 mm aluminum profiles, with other components made of aluminum alloy 6061. The motor selected is a worm gear reduction motor with a reduction ratio of 1:862 and a rated torque of 20 Nm. It is installed on the active gear train side, and power transmission is achieved through synchronous belts. The complete prototype of the robot is shown in Figure 8a. The guardrail inspection robot’s motor is powered by a 24 V, 1 A direct current, and its reciprocating motion—controlled by the forward and reverse rotation of the motor—is managed using two switches.

5.2. Robot Experimental Verification

Based on the simulation tests of the guardrail inspection robot described earlier, this experiment was conducted. Due to experimental limitations, a small-curvature curved guardrail section was not included. However, the simulation analysis in the previous section confirmed that the robot’s performance on small-curvature curved guardrail sections is similar to its performance on straight guardrail sections. Therefore, a straight guardrail section was used to represent the entire barrier-free guardrail section. The experimental results demonstrate that, in the barrier-free section, both the rubber-coated wheel and the auxiliary wheel meet the expected conditions, and the robot operates smoothly without detaching from the guardrail, as shown in Figure 8b.

The testing procedure for the robot’s obstacle-crossing performance in the barrier section is similar to that of the barrier-free section. The obstacle height and angle at the joint are adjusted by securing cardboard at the joint, transforming the barrier-free section into an obstacle section and simulating various obstacle-crossing scenarios. Through practical testing, it was found that the maximum obstacle height between the undamaged guardrails is 10 mm, and it cannot be adjusted to 12.5 mm. This confirms the guardrail inspection robot’s ability to navigate the normal guardrail conditions effectively.

On the obstacle guardrail section, the guardrail inspection robot encounters four obstacle-crossing scenarios: front auxiliary wheel obstacle crossing, rear auxiliary wheel obstacle crossing, non-driven rubber-coated wheel obstacle crossing, and driven rubber-coated wheel obstacle crossing, as illustrated in Figure 8d.

The guardrail section was installed at an approximate inclination of 3°, with the obstacle height at the junction of the two guardrail panels set to 10 mm, as shown in Figure 8c. The experiments revealed that the guardrail inspection robot exhibits excellent climbing capability, operating smoothly on the inclined guardrail section and successfully overcoming the obstacles even when the guardrail panels are tilted, meeting the design requirements.

It is important to note that, in real-world applications, the guardrails may be bent, cut, and reconnected during installation to adapt to the terrain conditions. The robot studied in this paper is currently not suitable for sections of guardrails with special connections or those deformed due to corrosion. Further research is needed to address these challenges.

6. Conclusions

This study proposes a mobile platform for guardrail inspection robots tailored to the working environment of highways, addressing the gap between the inspection requirements in road traffic management and the current capabilities of highway inspection robots. Structurally, an innovative adaptive double-sided wheel tensioning mechanism was designed. Through theoretical calculations, simulation analysis, and experimental testing, the robot demonstrated excellent operational and obstacle-crossing performance. The study also clarified the impact of specific parameter changes on the robot’s operation, providing a theoretical foundation for subsequent research. Furthermore, the design presented in this study can be extended to other types of highway guardrails and adapted for track inspection operations in other fields.