Assessing Infrastructure Readiness of Controlled-Access Roads in West Bangkok for Autonomous Vehicle Deployment

Abstract

The deployment of autonomous vehicles (AVs) depends on the readiness of both physical and digital infrastructure. However, existing national and city-level indices often overlook deficiencies along specific routes, particularly in developing contexts such as Thailand, where infrastructure conditions vary widely. This study develops and applies a corridor-level framework to assess AV readiness on five controlled-access roads in western Bangkok. The framework evaluates key infrastructure dimensions beyond conventional vehicle requirements. In this study, infrastructure readiness means the extent to which essential physical (EV charging capacity, traffic sign visibility, and lane marking retroreflectivity) and digital (5G speed and coverage) subsystems meet minimum operational thresholds required for AV deployment. Data were collected through field measurements and secondary sources, utilizing tools such as a retroreflectometer, a handheld spectrum analyzer, and the Ookla Speedtest application. The results reveal significant contrasts for physical infrastructure, showing that traffic signage is generally satisfactory, but EV charging capacity and road marking retroreflectivity are insufficient on most routes. On the digital side, 5G coverage was generally adequate, but network speeds remained less than half of the global benchmark. Kanchanaphisek Road demonstrated comparatively higher digital readiness, whereas Ratchaphruek Road exhibited the weakest road marking conditions. These findings point out the need for stepwise enhancements to EV charging infrastructure, lane marking maintenance, and digital connectivity to support safe and reliable AV operations. The proposed framework not only provides policymakers in Thailand with a practical tool for prioritizing corridor-level investments but also offers transferability to other rapidly developing urban regions experiencing similar infrastructure challenges for AV deployment.

1. Introduction

Autonomous vehicles (AVs) are considered a transformative technology capable of improving road safety, enhancing transport efficiency, and contributing to more sustainable mobility systems. Globally, progress toward AV deployment has been monitored through readiness indices at the national and city scales. While these indices provide valuable benchmarks, they often fail to capture the localized conditions that directly support AV operations. This limitation is particularly evident in developing regions, where infrastructure performance can be inconsistent across urban corridors.

Recent studies in both developed and developing contexts confirm that corridor-level assessments provide critical insights into operational readiness. For instance, the study in Leeds, UK, demonstrated how uneven road conditions restricted AV usability [1], while the study in China found that degraded lane marking retroreflectivity significantly impaired AV perception systems [2]. In developing regions, prior studies identified similar challenges, particularly in relation to EV charging infrastructure and localized digital coverage [3,4]. These examples indicate the need for sub-city or route-specific evaluations in contexts such as Bangkok.

Thailand exemplifies this challenge. Despite strong policy ambitions under the Thailand Smart Mobility Framework 2024, the country’s road network still faces heterogeneous traffic conditions, inconsistent enforcement, and uneven digital coverage. Moreover, empirical studies remain limited, particularly at the corridor level, leaving policymakers with little evidence on localized readiness. Previous research has been selective, narrowly focused, or dependent on expert opinion rather than systematic field data. As a result, critical indicators, such as EV charging capacity, road marking and traffic sign conditions, and 5G connectivity, are not well documented despite their direct implications for AV functionality.

To address this gap, this study introduces a multi-dimensional framework for assessing the readiness of controlled-access roads in western Bangkok. The proposed framework integrates physical infrastructure metrics (EV charging capacity, traffic sign visibility, and road marking retroreflectivity) with digital infrastructure indicators (5G network connectivity) to provide route-specific insights that complement existing city- and national-level indices. EV charging capacity is the effective number of EVs that can be reliably served per station per day, normalized by expected EV traffic along the corridor. Infrastructure readiness for controlled-access roads is the degree to which essential physical and digital subsystems meet operational thresholds necessary for AV deployment.

This study contributes to the AV readiness research by adapting global assessment principles to a route-level urban context and integrating physical and digital indicators into normalized scores. This offers critical insights for policymakers and planners to prepare local infrastructure for future AV deployment.

2. Background and Literature Review

2.1. Global Autonomous Vehicle Readiness Frameworks

Research on autonomous vehicle (AV) readiness has progressed from technology-oriented studies to more comprehensive frameworks that integrate policy, infrastructure, and societal factors. One of the most recognized tools, the autonomous vehicles readiness index (AVRI) developed by KPMG, has been used to evaluate countries across four domains, including policy and legislation, technology and innovation, infrastructure, and consumer acceptance [5]. The 2020 AVRI highlights global readiness status, with Singapore ranking highest due to proactive policies and infrastructure investments. The report also demonstrates a global trend of growing AV readiness, with approximately two-thirds of the 30 participating countries and jurisdictions designating or approving specific areas for AV testing.

Building on these country-level benchmarks, parallel research has adapted readiness frameworks to the urban scale. For example, a connected and autonomous vehicle (CAV) readiness index was developed for U.S. cities to identify disparities in preparedness across 52 jurisdictions [6]. Their assessments reflect uneven institutional capacity, infrastructural maturity, and technological adoption. Such city-focused frameworks provide a better understanding of readiness and highlight the challenge of translating national strategies into effective local implementation.

While these global and city-focused frameworks provide useful benchmarks, they often overlook contexts such as Thailand, where dense mixed traffic, limited AV legislation, and inconsistent infrastructure create distinctive challenges [7,8]. Addressing these localized conditions is critical to ensure that readiness assessments move beyond national averages and reflect the operational realities of specific corridors.

In addition to these aggregated approaches, recent studies have emphasized the critical role of infrastructure in supporting AV operations. For example, efforts were made to develop a CAV-readiness framework that guides road authorities in assessing physical and digital infrastructure for the cooperative intelligent transport systems (C-ITS) service support [9,10]. The framework emphasizes their role in enabling real-time vehicle communication and coordination. Similarly, a recent review on vehicular collaborative perception demonstrated that robust vehicle-to-infrastructure (V2I) and vehicle-to-vehicle (V2V) connectivity can enhance operational reliability and traffic safety [11]. The study by Luu et al. [12] offered a comprehensive review of digital infrastructure components essential to connected and automated vehicles. They categorized infrastructure into vehicle-based roadside, operational, and planning types. In the UK, there was an attempt to develop a corridor-level assessment framework based on expert judgment, and it pointed out that inconsistent infrastructure conditions can significantly restrict AV usability [1]. Findings by Babić et al. [2] reaffirmed that physical infrastructure characteristics such as the quality of lane markings, the clarity of signage, and the stability of digital connectivity are critical determinants of AV sensor performance and thus essential to functional readiness.

Building on this evidence, current practice guidelines and studies support corridor-scale diagnostics. Road-authority guidance recommends route-level audits of lane markings, signage, ITS assets, and the supporting digital infrastructure for CAV operation [13,14,15]. Recent studies show that lane marking retroreflectivity/contrast and their within-corridor variability are primary determinants of machine-vision detectability, which supports the need for corridor-specific performance thresholds rather than network averages [2,16,17].

Digital infrastructure readiness is likewise defined at the corridor scale. Cross-border CCAM trials in Europe specify and measure throughput, handover stability, and edge-computing availability along designated routes [18,19]. Recent policy also articulates the objective of uninterrupted 5G corridors on major transport routes, which support the use of network speed and coverage as corridor-level indicators in readiness assessments [20].

In developing countries, corridor-focused research remains limited but is emerging. The study in India emphasized that fragmented EV charging infrastructure constrains AV and EV adoption [3], while the study in Ecuador illustrated how corridor-level planning of fast-charging stations can directly influence deployment feasibility [4]. These studies highlight the growing recognition that route-level readiness is crucial, especially where national averages mask local deficiencies.

Given the convergence of automation and electrification, readiness assessments should integrate EV charging infrastructure and communications networks rather than treat them separately. Recent studies show that AV deployment is dependent on charging availability [21]. At the same time, operational readiness depends on both physical assets (e.g., lane marking quality and signage clarity) and digital performance indicator (e.g., 5G continuity) [2,22,23].

Despite these readiness research advancements, most existing readiness indices remain broad in scale, which privileges national or citywide averages and overlooks the variability that exists at the corridor or route level. A reliance on such aggregated measures could lead to localized gaps that may critically affect AV performance [24,25]. This limitation is particularly important in dense and mixed urban contexts, where differences in infrastructure quality, enforcement, and digital coverage can vastly influence the operational reliability of AVs. Addressing this analytical deficiency requires a shift toward corridor-level evaluations that capture the detailed environment in which AVs will operate. Building on this evidence, our study responds directly to this gap by applying a systematic, field-based framework to controlled-access roads in Bangkok. By integrating both physical and digital infrastructure indicators at the corridor scale, the study provides operationally relevant insights that national or city-level indices alone cannot reveal.

2.2. Infrastructure Readiness in Thailand

Thailand’s AV readiness challenges are inherently corridor-specific. Although global benchmarks have been established, they often overlook the contexts of developing countries [26], where diverse traffic conditions and inconsistent infrastructure remain absent from autonomous vehicle (AV) research [25]. For example, Bangkok’s road network is characterized by mixed traffic flows, inconsistent enforcement, high levels of pedestrian activity, and uneven digital coverage [8]. In addition, the absence of AV-specific legislation and technical standards also constrains structured pilot testing [27]. This situation points out the need for localized research and frameworks that can address these specific challenges, rather than relying on national-level or globally aggregated assessments. The integration of localized research is essential for informing targeted policy interventions and facilitating successful AV adoption in the region.

Existing AV readiness research in Thailand has been fragmented or narrowly focused. For example, a recent study conducted a preliminary infrastructure assessment through interviews with experts and reported that the country is not fully prepared for AV adoption [7]. The study identified critical challenges in both physical and digital infrastructure. The study revealed that private roads appeared more AV-ready than public roads, which often lacked intelligent transportation system (ITS) facilities, 5G connectivity, and vehicle-to-everything (V2X) capabilities. In addition, the absence of a clear policy framework and limited government investment were cited as major barriers. However, the study did not provide a comprehensive and quantitative evaluation of AV readiness.

Another study in Thailand examined highway upgrades in the Eastern Economic Corridor (EEC) through a physical infrastructure readiness assessment [28]. The study utilized field investigation and accident data as key components for the assessment. The study recommended low-cost improvements such as enhanced lane markings, intelligent cameras, and energy-efficient lighting. The study findings suggested that significant readiness improvements could be achieved without major overhauls.

2.3. Relation to Existing Research

Global indices such as the AVRI [5] and city-focused CAV readiness indices [6] emphasize the importance of infrastructure quality, digital connectivity, and regulatory preparedness. However, their national or city-scale assessments are insufficient for identifying deficiencies along specific corridors. Prior studies have highlighted that inadequate EV charging capacity can constrain AV and EV adoption in countries such as India and Ecuador [3,4], while degraded road markings have been shown to undermine AV perception systems in China [2]. Likewise, uneven sub-city 5G network performance has been reported as a barrier to continuous V2X communication [24]. This study addresses these gaps by contributing a route-level assessment that captures intra-urban variation and integrates both physical and digital infrastructure indicators into a unified readiness framework.

3. Methodology

3.1. Conceptual Framework

The proposed framework evaluates AV readiness through two dimensions, including physical and digital infrastructure. These scopes were selected to capture the distinct requirements of AV operation beyond conventional vehicles, such as reliable lane detection, uninterrupted connectivity, and adequate EV support facilities.

To enable comparability across indicators with different measurement units, all results were normalized to a 10-point scale. This ensures that metrics such as 5G speed (Mbps), retroreflectivity (mcd·lx⁻¹·m⁻²), and EVs served per charger can be integrated without bias toward their absolute magnitude. A uniform scale also allows policymakers to intuitively interpret readiness levels. In this study, no additional weighting was applied, and each indicator was treated equally to maintain methodological transparency and reproducibility. However, the framework is designed to accommodate weighted schemes in future applications, where context-specific priorities, such as emphasizing digital connectivity in highly automated environments, may warrant differential importance.

3.1.1. Physical Road Infrastructure

Three elements of physical infrastructure were assessed:

• Electric charging infrastructure was evaluated through a capacity-based metric that considers both the supply of chargers and the estimated demand from registered EV penetration.
• Traffic sign visibility was measured through field surveys using a four-point rating system. Signs were scored on clarity, legibility, and obstruction, with higher scores reflecting optimal conditions for AV sensor recognition.
• Road marking retroreflectivity was tested with a retroreflectometer at regular intervals. Measurements followed the standards of Thailand’s Department of Highways, with thresholds of 150 mcd·lx⁻¹·m⁻² for acceptable reflectivity. These data provide an indication of whether road markings remain detectable under nighttime or low-visibility conditions.

3.1.2. Digital Road Infrastructure

Digital infrastructure was assessed in terms of both network performance and coverage:

• 5G network speed was measured using the Ookla Speedtest application at one-kilometer intervals while vehicles traveled at operational speeds. The results were benchmarked against Qatar, which has the world’s highest reported 5G speed as of March 2024 [29].
• 5G network coverage was captured through a handheld spectrum analyzer, which records signal strength at 200 m intervals along each route. The percentage of continuous mid-band 5G coverage was calculated as a measure of digital reliability.

Through the integration of these indicators, the framework extends beyond conventional readiness indices by offering a more comprehensive assessment that captures both physical and digital conditions at the corridor scale. In this study, infrastructure readiness on controlled-access roads is defined as the degree to which essential physical and digital subsystems meet operational thresholds necessary for autonomous vehicle deployment. Physical readiness includes EV charging support, signage legibility, and lane marking retroreflectivity, while digital readiness reflects 5G connectivity and speed. Together, these components form a multidimensional readiness profile that complements national and city-level indices by capturing corridor-specific conditions. In addition, EV charging capacity is incorporated not only as an indicator of electrification progress but also as a prerequisite for autonomous fleets, which are expected to rely on electric power. Thus, the concept of readiness presented here integrates the energy dimension (EV charging) with road infrastructure (signs and markings) and communication systems (5G connectivity), enabling a comprehensive evaluation of corridor-level preparedness for AV operations.

3.2. Study Area

The study focused on five controlled-access roads in western Bangkok:

• Borommaratchachonnani Road (17.6 km)
• Elevated Borommaratchachonnani Road (18 km)
• Ratchaphruek Road (10.6 km)
• Rama 2 Road (14.7 km)
• Kanchanaphisek Road (30.3 km)

These roads serve as major corridors that link the western area to central Bangkok. They were selected due to their strategic role in urban mobility and their potential to serve as early deployment sites for AV operations. The locations are shown in Figure 1.

3.3. Data Collection and Assessment Methodology

3.3.1. Physical Infrastructure

EV charging capacity in this framework is defined as the effective number of EVs that can be reliably served per station per day, which is normalized against expected EV traffic volumes along each corridor. This composite metric integrates supply-side availability (e.g., station throughput based on charger type, number of ports, and operational hours) with demand-side usage (e.g., expected EV penetration derived from average daily traffic and EV registration ratios). By balancing available charging supply with the projected usage, this indicator highlights potential capacity deficits that could constrain electric and autonomous vehicle adoption and corridor-level readiness.

• EV Charging Capacity

The evaluation of EV charging infrastructure for individual roads is based on the EV charging capacity score, which quantifies the amount of available charging capacity per EV at a charging station. This metric requires an estimation approach that reflects the actual usage patterns of EVs along the study roadway segments. The EV charging capacity score is defined as

$$\mathrm{E}\mathrm{V} \mathrm{C}\mathrm{h}\mathrm{a}\mathrm{r}\mathrm{g}\mathrm{i}\mathrm{n}\mathrm{g} \mathrm{C}\mathrm{a}\mathrm{p}\mathrm{a}\mathrm{c}\mathrm{i}\mathrm{t}\mathrm{y} \mathrm{S}\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{e}=\left({\displaystyle \frac{\mathrm{C}\mathrm{h}\mathrm{a}\mathrm{r}\mathrm{g}\mathrm{i}\mathrm{n}\mathrm{g} \mathrm{C}\mathrm{a}\mathrm{p}\mathrm{a}\mathrm{c}\mathrm{i}\mathrm{t}\mathrm{y}}{\mathrm{E}\mathrm{V}\mathrm{s} \mathrm{p}\mathrm{e}\mathrm{r} \mathrm{C}\mathrm{h}\mathrm{a}\mathrm{r}\mathrm{g}\mathrm{i}\mathrm{n}\mathrm{g} \mathrm{S}\mathrm{t}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}}\right)\times 10$$

where

○

Charging capacity refers to the number of EVs that can be served by a single charging station per day. For two dual-port chargers operating 12 h per day with an average charging time per port of 30 min per EV [30], the charging capacity is calculated as

12 h per day × 2 EVs charging per hour per port × 2 ports per charger × 2 chargers per station = 96 EVs charging/day/station

EVs per charging station is estimated as the ratio between the expected number of EVs on the study roadway segments and the total number of charging stations within the study area. The expected number of EVs along the studied roads is calculated as

$$\mathrm{E}\mathrm{x}\mathrm{p}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{e}\mathrm{d} \mathrm{N}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r} \mathrm{o}\mathrm{f} \mathrm{E}\mathrm{V}\mathrm{s}=\mathrm{R}\mathrm{o}\mathrm{a}\mathrm{d}\mathrm{w}\mathrm{a}\mathrm{y} \mathrm{A}\mathrm{D}\mathrm{T}\times 1.9\%$$

where

○

Roadway ADT represents the average daily traffic volume, obtained from traffic data provided by the Department of Highways [31] and the Department of Rural Roads [32].

○

The constant (1.9%) denotes the proportion of registered EVs relative to the total vehicle population, based on cumulative vehicle registration data for the Bangkok metropolitan area as of December 2024 [33].

This method enables the assessment of the adequacy of EV charging infrastructure by accounting for both supply-side capacity and demand-side EV presence along the study roadway segments.

2.

Traffic Sign Visibility

In assessing the visibility of the traffic signs on the study roads, a four-point scoring system was used based on the descriptions shown in

Table 1. Traffic sign score system.

Rating Score	Descriptions
1	The traffic sign has low clarity, making it difficult to read, and there are obstructions blocking it.
2	The traffic sign has low clarity, making it difficult to read, with no obstructions blocking it.
3	The traffic sign is clear and visible, but there are minor obstructions blocking it.
4	The traffic sign is clear and visible, with no obstructions blocking it.

, where a score of 4 indicates the highest visibility conditions for drivers, while a score of 1 represents the lowest visibility. Three types of traffic signs were considered in the evaluation: warning signs, control signs, and navigation signs. The data were collected through a visual inspection of the physical condition of the signs. Once the scores of each sign type were obtained, the average score was then calculated for each of the study roads.

3.

Road Marking Retroreflectivity

The retroreflectivity of road markings was measured in April 2024. The measurement was performed using a retroreflectometer unit to measure nighttime visibility from 15 m (RL15). Measurements were collected every 500 m along the study road sections. The retroreflectivity data were then analyzed and compared to the criteria used by Thailand’s Department of Highways, which specifies that the retroreflectivity of white road markings should be no less than 300 mcd·lx⁻¹·m⁻² when newly painted and 150 mcd·lx⁻¹·m⁻² after 3 months. Therefore, the retroreflectivity of road markings is considered acceptable if it is greater than 150 mcd·lx⁻¹·m⁻², as shown in

Table 2. Criteria for Evaluating the Retroreflectivity of Road Markings.

Measurement Type	Criteria (mcd·lx−1·m−2)
	Not Acceptable	Acceptable
RL15	0–149	150+

. The assessment of road markings used in this study is based on a calculation of the retroreflectivity percentage that passes the criteria for each route.

3.3.2. Digital Infrastructure

• Network Speed

The process of collecting wireless connectivity speed through 5G networks on mobile devices entails gathering signals from the three largest mobile networks in Thailand. Network speed data were collected through the Speedtest application by Ookla [34]. The data were collected for every 1 km while the vehicle was traveling at a speed between 60 and 80 km per hour. After that, the maximum and average network speeds were calculated for each route. The assessment of the network speed was performed by calculating the percentage of the resulting network speed compared to the world benchmark, which is 313.3 Mbps in Qatar as of March 2024.

2.

Network Coverage

The data for 5G coverage was collected using the handheld spectrum analyzer (Model N9340B; Keysight Technologies, Santa Rosa, CA, USA) as shown in Figure 2. The measurement procedure requires the following steps. First, the analyzer is set up to identify signals in the frequency band between 1 GHz and 35 GHz, which is considered the low and high bands of 5G technology. The data were recorded at intervals of 200 m along the route while the vehicle was traveling at speeds between 60 and 80 km per hour. The 5G network coverage is based on mid-band frequency ranges between 1.7 GHz and 2.6 GHz. The assessment of the 5G network coverage was then performed by calculating the percentage of 5G coverage along each route.

To facilitate comparison, all results were normalized to a 10-point scale. This enabled integration of different indicators into a consistent readiness score for each corridor, which can support comparative analysis of both physical and digital dimensions.

4. Results

4.1. Assessment Results for EV Charging Infrastructure

Table 3. Assessment results for readiness of EV charging infrastructure.

Study Roads	Number of Roadside Charging Stations	ADT	Expected EVs (ADT × 1.9%)	EVs Per Charging Station	EV Charging Capacity Score
Road 1	25	1,297,278.0	2465.0	98.6	9.7
Road 2 *	5	69,844.0	1327.0	265.4	3.6
Road 3	16	136,447.0	2593.0	162.0	5.9
Road 4	24	260,090.0	4942.0	205.9	4.7
Road 5	21	167,210.0	3177.0	151.3	6.3

Road 1 = Borommaratchachonnani Road, Road 2 = Elevated Borommaratchachonnani Road, Road 3 = Ratchaphruek Road, Road 4 = Rama 2 Road, and Road 5 = Kanchanaphisek Road. * The Elevated Borommaratchachonnani Road has access to the charging stations on Borommaratchachonnani Road through their connecting ramps.

presents the assessment results of EV charging infrastructure readiness based on the EV charging capacity score. Borommaratchachonnani Road achieved the highest score at 9.7, followed by Kanchanaphisek Road and Ratchaphruek Road, with scores of 6.3 and 5.9, respectively. Rama 2 Road and Elevated Borommaratchachonnani Road recorded the last two lowest scores, at 4.7 and 3.6, respectively.

As illustrated in Figure 3, EV charging capacity remains the weakest dimension of physical infrastructure readiness across all routes. While Borommaratchachonnani Road demonstrates relatively stronger performance, none of the study roads provides sufficient charging support for large-scale AV adoption.

4.2. Assessment Results for Traffic Signs

The assessment results for the readiness related to traffic signs are shown in

Table 4. Assessment results for readiness of traffic signs.

Study Roads	Warning Type		Control Type		Navigation Type		Total No. of Signs	Average Score
	No. of Signs	Score	No. of Signs	Score	No. of Signs	Score
Road 1	9	3.6	4	3.0	10	4.0	23	3.5
Road 2	23	4.0	7	3.4	22	3.5	52	3.6
Road 3	18	4.0	4	3.5	18	3.3	40	3.6
Road 4	66	4.0	7	4.0	22	3.6	95	3.9
Road 5	30	3.7	22	3.4	35	3.8	87	3.6

Road 1 = Borommaratchachonnani Road, Road 2 = Elevated Borommaratchachonnani Road, Road 3 = Ratchaphruek Road, Road 4 = Rama 2 Road, and Road 5 = Kanchanaphisek Road.

. It is found that all of the study roads achieved relatively high average scores in the range of 3.5 to 3.9 out of 4.0. The highest average score of 3.9 was obtained from Rama 2 Road, followed by a tied score of 3.6 for the elevated Borommaratchachonnani Road, Ratchaphruek Road, and Kanchanaphisek Road, and the lowest score of 3.5 for Borommaratchachonnani Road. In general, the traffic signs on the study roads are well maintained; their visibility is not impaired and only slightly obscured in some areas.

As shown in Figure 4, traffic sign visibility is generally strong across all routes. Average scores above 3.5 indicate that signage is not a critical barrier to AV sensor recognition in the study area.

4.3. Assessment Results for Road Markings

The retroreflectivity of road markings across the study routes was assessed to determine their readiness for detection by autonomous vehicles.

Table 5. Assessment Results for Readiness of Road Markings.

Study Roads	Average (mcd·lx−1·m−2)	Standard Deviation (mcd·lx−1·m−2)	Percentage of Road Markings That Meet the Standard (%)
Road 1	117.1	72.1	28.6
Road 2	158.6	134.5	33.3
Road 3	81.7	6.8	0.0
Road 4	204.7	193.2	33.3
Road 5	136.1	114.5	37.5

Road 1 = Borommaratchachonnani Road, Road 2 = Elevated Borommaratchachonnani Road, Road 3 = Ratchaphruek Road, Road 4 = Rama 2 Road, and Road 5 = Kanchanaphisek Road.

summarizes these findings by showing the percentage of the route distance that meets the Department of Highways’ standards. Overall, the results indicate that a majority of the road markings do not meet the acceptable standard for reliable AV navigation. The highest compliance was observed on Kanchanaphisek Road, with 37.5% of the total distance meeting the standard. This rate was followed by the elevated Borommaratchachonnani Road and Rama 2 Road, both at 33.3%. Borommaratchachonnani Road had 28.6% meeting the standard, while most of the road markings on Ratchaphruek Road have started to fade, with 0% meeting the standard.

The average retroreflectivity values across all routes ranged from 81.7 mcd·lx⁻¹·m⁻² on Ratchaphruek Road to 204.7 mcd·lx⁻¹·m⁻² on Rama 2 Road. The low value on Ratchaphruek Road indicates a degradation of the markings. Variability in retroreflectivity was also examined. Ratchaphruek Road showed the most consistent readings with a low standard deviation of 6.8 mcd·lx⁻¹·m⁻², while Rama 2 Road, the elevated Borommaratchachonnani Road, and Kanchanaphisek Road exhibited the highest variability, with standard deviations of 193.2, 134.5, and 114.5 mcd·lx⁻¹·m⁻², respectively. The high variability on these routes suggests inconsistent maintenance or localized degradation, which could pose a safety concern for AVs.

As presented in Figure 5, compliance with retroreflectivity standards is uneven and often insufficient. These deficiencies suggest that faded or poorly maintained markings may compromise AV detection capabilities, particularly at night or under low-visibility conditions.

4.4. Assessment of 5G Network Speed

In assessing the readiness of 5G network speed to support autonomous vehicles, data from Ookla was used to measure mobile network speeds of the three major networks along the study roads and compared to the network speed in Qatar, which has a record of the global top network speed at 313.3 Mbps. From the results in

Table 6. Assessment Results for 5G Network Speed.

Study Roads	Assessment of 5G Network Speed
	Speed (Mbps)					% of World Benchmark
	Net Work 1	Net Work 2	Net Work 3	Avg.	Std.	Network 1	Network 2	Network 3	Avg.
Road 1	62.9	31.3	62.2	52.1	41.3	20.1	40.0	19.8	16.6
Road 2	43.4	64.9	138.0	82.1	62.0	13.9	20.7	44.0	26.2
Road 3	162.3	39.6	154.5	118.8	87.3	51.8	12.6	49.3	37.9
Road 4	131.7	39.0	119.7	96.8	83.1	42.0	12.4	38.2	30.9
Road 5	192.9	70.9	177.1	147.0	113.9	61.6	22.6	56.5	46.9

Road 1 = Borommaratchachonnani Road, Road 2 = Elevated Borommaratchachonnani Road, Road 3 = Ratchaphruek Road, Road 4 = Rama 2 Road, and Road 5 = Kanchanaphisek Road. Avg. = Average, Std. = Standard Deviation.

, it is found that the 5G network speed on all the study routes was in a range between 52.1 and 147.0 Mbps. Kanchanaphisek Road had the highest overall average speed at 147.0 Mbps, which is 46.9% of the world benchmark. Borommaratchachonnani Road had the lowest network speed (52.1 Mbps) and the lowest percentage (16.6%) of the global best speed. The results indicate that the overall 5G network speed for all the study roads was still less than half of the world benchmark. Overall, 5G speeds show high variability across all roads, especially on those with diverse usage patterns or complex physical characteristics. This high variability indicates that while some locations have excellent signal quality, others experience significantly lower performance. Kanchanaphisek Road may be suitable for applications requiring high-throughput connectivity, such as autonomous vehicles.

As illustrated in Figure 6, average 5G speeds for each route remain less than half of the world benchmark on all corridors, with large variability across routes. This indicates inconsistent digital readiness, which could challenge reliable V2X communication required for autonomous operations.

4.5. Assessment Results for 5G Network Coverage

Figure 7 shows the assessment results for 5G network coverage on all the study roads. Kanchanaphisek Road had the best network coverage at 97.5% of the entire study route, followed by Borommaratchachonnani Road, with a coverage of 84.1%. The field data collection reveals that both Kanchanaphisek and Borommaratchachonnani Roads had less interruption for a reception of midband frequencies used in 5G networks compared to the other three roads, which had coverage in a range between 71.2% and 78%. The coverage of Rama 2 Road was only 71.2% due to ongoing roadwork constructions that can hinder the signal reception on some portions of the road.

4.6. Summary of Assessment Results for All Categories

Table 7. Normalized scores for physical and digital infrastructure readiness.

Study Roads	Physical Infrastructure			Digital Infrastructure
	EVs Charging Capacity Score (10-Based) a	Traffic Signs Score 4-Based (10-Based) b	Percentage of Road Markings That Meet the Criteria % (10-Based) c	5G Network Speed % (10-Based) d	5G Coverage % (10-Based) e
Road 1	(9.7)	3.5 (8.8)	28.6 (2.9)	16.6 (1.7)	84.1 (8.4)
Road 2	(3.6)	3.6 (9.0)	33.3 (3.3)	26.2 (2.6)	74.0 (7.4)
Road 3	(5.9)	3.6 (9.0)	0.0 (0.0)	37.9 (3.8)	78.0 (7.8)
Road 4	(4.7)	3.9 (9.8)	33.3 (3.3)	30.9 (3.1)	71.2 (7.1)
Road 5	(6.3)	3.6 (9.0)	37.5 (3.8)	46.9 (4.7)	97.5 (9.8)

Road 1 = Borommaratchachonnani Road, Road 2 = Elevated Borommaratchachonnani Road, Road 3 = Ratchaphruek Road, Road 4 = Rama 2 Road, and Road 5 = Kanchanaphisek Road. Note: ^a EVs Charging Capacity Score (10-based) = Available charging capacity per EV charging station from Table 3. ^b Traffic Sign Score (10-based) = Traffic sign score (4-based) from Table 4 multiplied by 2.5. ^c Percentage of Road Markings that Meet the Criteria (10-based) = Percentage of Road Markings that Meet the Criteria (%) from Table 5 divided by 10. ^d 5G Network Speed (10-based) = 5G Network Speed (% of World Benchmark) from Table 6 divided by 10. ^e 5G Coverage (10-based) = 5G Coverage (% of Route Distance) from Figure 7 divided by 10.

presents the normalized scores for physical and digital infrastructure, converted to a 10-point scale for consistency in comparison and evaluation.

Figure 8 presents a summary of the assessment results regarding the physical and digital infrastructure readiness of major roads in western Bangkok. All components were evaluated on a scale of 0 to 10, with the following key findings.

• EV charging infrastructure shows some deficiencies for several corridors, with the highest score observed on Borommaratchachonnani Road at 9.7 out of 10.
• Traffic signs are generally well-maintained, with a score of 8.8 or higher.
• Road markings meet the required retroreflectivity standards on less than 50% of the assessed segments, indicating detection challenges for autonomous vehicles.
• 5G network speed falls below 50% of the global benchmark on all roads, with Kanchanaphisek Road recording the highest average speed of 147.0 Mbps, which is equivalent to a readiness score of 4.7.
• 5G network coverage is highest on Kanchanaphisek Road, with a score of 9.8, followed by Borommaratchachonnani Road, with a score of 8.4, while the remaining roads range from 7.1 to 7.8.

5. Discussion

This study proposes a framework for assessing the readiness of both physical and digital infrastructure along controlled-access roads. The framework combines international benchmarks with localized indicators and was applied to five major corridors in western Bangkok. The framework provides practical, route-level insights into current conditions that are directly relevant to autonomous vehicle (AV) deployment. Existing research has not sufficiently engaged with this analytical scale, leaving a gap between broad national or city indices and the localized conditions that AVs encounter in practice.

5.1. Interpretation of Results

The assessment results indicate corridor-level variations that reveal critical infrastructure gaps for autonomous vehicle (AV) deployment in Bangkok. Among the physical infrastructure indicators, EV charging capacity emerged as the weakest component across all corridors. Even on Borommaratchachonnani Road, which achieved the highest score, the capacity remains only moderately adequate compared to the anticipated demand. This deficiency aligns with broader regional patterns, such as those observed in India and Ecuador, where limited charging infrastructure has similarly hindered the adoption of electric and autonomous vehicles [35]. This limitation poses a direct barrier to scaling AV operations, which are expected to rely predominantly on electrified propulsion systems.

By contrast, traffic signage across the study corridors was consistently strong. Average scores exceeded 3.5 out of 4.0, indicating that most signs are legible, unobstructed, and suitable for detection by AV sensors. However, this relative strength cannot compensate for the substantial weaknesses observed in other areas. The analysis revealed that less than half of the study road segments met the minimum retroreflectivity thresholds, with some corridors, such as Ratchaphruek Road, recording below-standard values. These outcomes point to critical areas for enhancement to ensure reliable AV lane-detection capabilities. In real driving conditions, degraded or faded markings can cause AV perception systems to misinterpret lane boundaries, leading to unsafe lane departures or forcing disengagements where the human driver must resume control. Such failures are most critical at night or during rainfall, when reliance on retroreflectivity is highest [36]. These findings mirror recent evidence from China, where high-quality markings are emphasized as pivotal for reliable perception.

Digital infrastructure performance was similarly uneven. While 5G coverage was relatively strong on Kanchanaphisek and Borommaratchachonnani Roads, average network speeds across all corridors remained less than half of the global benchmark. This variation is consistent with prior observations of sub-city differences in network reliability and poses significant constraints for continuous vehicle-to-everything (V2X) communication. For AVs, weak or inconsistent 5G signals translate into delayed hazard warnings, interruptions in platooning, and reduced reliability of cooperative maneuvers such as merging at ramps or coordinating at intersections. Even momentary latency can increase the risk of rear-end collisions or erratic driving responses in dense traffic [21,22]. For advanced automation, where cooperative maneuvers and real-time updates are essential, such variability represents a critical operational risk.

Overall, the results emphasize that AV readiness is not a single-dimensional measure but an integrated set of conditions encompassing propulsion, perception, and communication systems. While Thailand has strong national ambitions under the smart mobility framework, localized infrastructure gaps remain a barrier to reliable deployment. Kanchanaphisek Road demonstrated comparatively higher readiness, suggesting its potential as a pilot deployment corridor. However, achieving wider-scale AV adoption will require upgrades in EV charging provision, consistent lane marking maintenance, and 5G performance to establish resilient and safe operational environments for autonomous mobility.

5.2. Research Contributions

This study advances existing readiness indices by translating global assessment concepts into a multi-dimensional, normalized scoring system tailored to the corridor level. Unlike conventional indices, which are designed for cross-national comparisons, the proposed framework captures localized variation in road conditions, signage, and network performance. This finer resolution allows for a more accurate understanding of the operational environments that autonomous vehicles (AVs) will encounter in practice.

A key contribution of this research lies in its integration of physical and digital indicators into a single evaluative structure. By unifying metrics for EV charging capacity, road markings, traffic signs, and 5G connectivity, the framework moves beyond aggregated national or city-level assessments to provide actionable, route-specific insights. Such a comprehensive approach supports not only comparative analysis across corridors but also prioritization of targeted investments where deficiencies are most acute.

Furthermore, this study conceptualizes EV charging capacity as more than an electrification readiness. EV charging infrastructure is framed as a prerequisite for autonomous fleets, which are expected to rely predominantly on electric propulsion. This reframing emphasizes that infrastructure readiness is not limited to isolated components but encompasses the combination of physical attributes (charging, markings, signage) and digital connectivity. By incorporating both propulsion requirements and perception/communication systems, the framework provides a comprehensive basis for evaluating AV readiness and guiding infrastructure planning.

Beyond its immediate application in Bangkok, the framework is transferable to other urban and regional contexts because it relies on indicators that are universally measurable and operationally relevant, such as EV charging infrastructure capacity, signage legibility, lane marking retroreflectivity, and 5G connectivity. These parameters are not country-specific but represent fundamental conditions for AV functionality. Furthermore, the framework is scalable, as its modular structure allows for additional dimensions (e.g., weather resilience, cybersecurity readiness, or cooperative V2X penetration) to be incorporated without altering the baseline methodology. This flexibility ensures that the framework can evolve alongside technological advancements while maintaining comparability across corridors and cities.

5.3. Implications for Practice and Policy

The findings indicate the limitations of relying solely on national-level readiness scores, which may fail to identify where localized measures are most needed. Corridor-level diagnostics, as demonstrated in this study, provide a finer resolution of infrastructure performance that can directly inform practical interventions. By focusing on controlled-access roads in Bangkok, the framework highlights how differences in charging availability, marking conditions, and digital connectivity can shape the feasibility of autonomous vehicle deployment.

In terms of implementation, the results suggest that infrastructure upgrades should be prioritized according to their relative impact on AV operational reliability. In the short term, expanding EV charging capacity is critical, as insufficient availability directly constrains both electrification and automation [21,35]. Equally important is the establishment of systematic maintenance programs for road markings, since faded or inconsistent markings undermine AV perception systems and increase disengagement risks, particularly under low-visibility conditions [36]. Over the medium term, corridor-level investments in digital infrastructure, particularly enhancing 5G speed and reducing variability, are necessary to enable reliable vehicle-to-everything (V2X) communication. Weak or inconsistent signals can lead to delayed hazard warnings and elevated crash risks [21,22].

These priorities align with Thailand’s Smart Mobility Framework 2024, which emphasizes integration of electrification and digitalization. Pilot deployment along partially prepared corridors, such as Kanchanaphisek Road, could serve as testbeds for scaling nationwide readiness. Comparative experiences from Singapore, which ranks highest in AV readiness due to proactive policies and infrastructure investment [5], and India, where insufficient corridor-level EV charging facilities remain a major barrier [3], suggest that a phased, corridor-based improvement approach can optimize resource allocation while supporting faster AV adoption.

5.4. Study Limitations and Future Directions

While the study provides a detailed corridor-level perspective, its scope is limited to western Bangkok. Applying the framework to other areas within the metropolitan region and to additional cities would help identify broader infrastructure gaps and inform regional planning. In addition, time-dependent factors were not captured in this study. Seasonal variation, such as heavy rainfall during the monsoon season, may affect lane marking visibility, retroreflectivity, and even wireless signal performance. Similarly, time-of-day conditions such as nighttime driving or peak-hour congestion may influence both physical infrastructure performance and digital connectivity. Incorporating temporal data would capture seasonal and maintenance-related variation in road and network conditions, providing a more dynamic picture of readiness.

6. Conclusions

This study evaluates the readiness of corridor-level infrastructure for autonomous vehicle (AV) deployment along five controlled-access roads in western Bangkok. The framework integrates both physical and digital indicators to produce route-specific insights that complement national and city-scale assessments.

The findings reveal critical deficiencies that may impede AV operations, including insufficient EV charging facilities, poor road markings, and limited 5G connectivity. These limitations suggest the need for improvements at the corridor level, where deficient infrastructure has the greatest impact on operational reliability.

The study also refines global benchmarking methods into a multi-dimensional scoring system suited to corridor application. This approach supports comparison across infrastructure types and offers transferability for urban areas with uneven or inconsistent infrastructure conditions.

From a policy perspective, focusing upgrades along major corridors, particularly in EV charging capacity, lane marking conditions, and digital connectivity, can accelerate AV adoption and ensure more efficient allocation of investment.

Future research should expand the geographic scope and incorporate a systematic monitoring of infrastructure conditions over time to capture spatial variation and temporal trends. Attention should also be given to temporal factors such as seasonal weather patterns and time-of-day variations, which may significantly influence both the physical visibility of infrastructure (e.g., road markings under rainfall or nighttime conditions) and the stability of digital connectivity. This will strengthen the alignment between infrastructure development and AV deployment.