Science and Engineering Connect

A Study of Pedestrian Crossing Behaviors on Major Urban Roads to Inform Road Safety Engineering Design

2025-12-09T13:58:17+07:00

Background and Objectives: Pedestrian safety is widely recognized as a critical public health and urban transport concern, particularly in rapidly growing urban areas of developing countries where pedestrian injury and fatality rates remain disproportionately high. Thailand faces significant challenges in mitigating these pedestrian safety risks. Taking Khon Kaen, which is a major regional transit node undergoing rapid expansion and vehicular growth, as a case study, the present research addresses the intensified conflicts between pedestrians and vehicles, particularly along major arterials with substandard crossing infrastructure. Although local agencies have attempted to implement safety measures, empirical studies remain limited regarding validated behavioral measurement tools that integrate psychological and context-specific environmental factors within the Thai urban context. To address this gap, the present study developed a pedestrian behavior measurement scale and investigated the socio-environmental, legal, and demographic factors influencing road-crossing decisions in urban Khon Kaen. The study indeed aimed to 1) develop and examine the construct validity of a pedestrian behavior measurement scale; and 2) investigate the relationship between pedestrian behavior factors and social, environmental, legal enforcement, and general variables (age, gender, occupation, and walking distance) on major roads in the urban areas of Khon Kaen.

Methodology: We conducted a quantitative survey among urban residents of Khon Kaen. We developed a 15-item questionnaire informed by behavioral theories, prior pedestrian safety studies, and observations of local traffic conditions. The questionnaire addressed attitudes toward crossing, risk-taking behaviors, safety awareness, and environmental concerns, including exposure to air pollution during road crossing. 400 participants completed the survey (N=400), including students, workers, and the public, selected through random sampling. We used descriptive statistics for preliminary analysis; this was followed by Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA). EFA was performed using orthogonal (Varimax) rotation, with acceptance criteria including Bartlett’s Test of Sphericity (p < 0.05), Kaiser-Meyer-Olkin Measure (KMO > 0.6), and Cronbach’s alpha for reliability. CFA was conducted to evaluate model fit using such indices as CFI, GFI, RMR, and CMIN/DF.

Main Results: The results reveal that the population exhibited attitudes related to safe pedestrian crossings and had frequently encountered air pollution while crossing the streets. Additionally, the EFA identified two new factors (KMO = 0.735, Sig. = 0.000) viz. Society and Environment, and Violations. The CFA indicated that the model demonstrated a good fit with the empirical data (CMIN/DF = 1.890, GFI = 0.921, CFI = 0.901, RMR = 0.089). Gender and individual pedestrian behaviors were found to be significantly associated with the Society and Environment factor (p < 0.005). Age was significantly related to the Violations factor (p < 0.005).

Conclusions: Based on our analysis of 9 questionnaire items, two distinct behavioral factors were identified: (1) Society and Environment, and (2) Violations. These factors were significantly associated with participants’ gender, age group, and individual pedestrian behaviors. The study demonstrates that pedestrian behavior in Khon Kaen’s complex urban environment is shaped by both environmental conditions and personal decision-making patterns. The questionnaire provides a systematic tool for measuring pedestrians’ perceptions, risk awareness, and crossing behaviors. Together, the two latent factors capture how environmental influences, risky actions, and safety-oriented practices interact to shape pedestrian behavior. These empirical findings offer important insights that can inform traffic engineering strategies, guide urban management decisions, and support efforts to cultivate a stronger culture of pedestrian safety, especially in a similar context to that of Khon Kaen.

Practical Application: Our findings can inform the development of public safety campaigns and the design of pedestrian infrastructure, including raised crosswalks, pedestrian refuge islands, pedestrian signal systems, and improved street lighting. The results also support targeted enforcement efforts aimed at reducing risky crossing behaviors. By using pedestrians’ perceptions and behavioral data as a foundation for urban design, planners and policymakers can enhance the safety and walkability of road systems in Khon Kaen and other cities in the future.

Spatial-Spectral Deep Transfer Learning Network for Few-Shot Hyperspectral Image Classification

2026-01-12T16:33:09+07:00

Background and Objectives: In recent decades, classification of hyperspectral images (HSIs) has made significant progress with the development of deep learning (DL). However, learning with a few samples remains a difficulty due to the high cost of annotating samples and potential errors in manual evaluation. Most DL-related methods require a sufficient number of annotated samples to learn the intricate spatial-spectral structure and spectrum of HSI data, making them unsuitable for practical situations with limited samples. To overcome the difficulty in learning with a few samples, we propose a deep transfer contrastive learning approach known as Spatial–Spectral Deep Transfer Learning Network (SSDTLN) for HSI classification. The proposed approach utilizes a spectral augmentation process to increase the diversity of pairs of samples with diverse characteristics. In addition, a spatial feature extraction component is incorporated to leverage complementary information by supporting the spatial component with pretrained representations from the source domain and the spectral residual component, focusing on precise spectral feature extraction. Moreover, contrastive learning can increase the discriminative ability of the representations with the goal of improving the ability to deal with tasks concerning high similarities between classes and large variances within classes with few samples.

Methodology: The proposed approach involved a spatial-spectral few-shot deep transfer learning approach for the classification of hyperspectral images, which was broken down into three broad phases: spatial feature extraction via a CNN-based spatial network pre-trained on the ImageNet dataset; spectral feature extraction via a residual convolutional network coupled with spectral enhancement techniques such as random spectral shift and Gaussian noise injection; spatial-spectral feature fusion via convolutional layers; and a spatial attention module (SAM). Hyperspectral image patches were then created and arranged in a C-way K-shot episode, where the labeled images comprised the support set, while the remaining comprised the query set. A mapping layer was applied for projecting the high-dimensional hyperspectral image data into a three-channel format, compatible with the spatial network. Finally, the classification was conducted via a nearest-neighbor classifier, while the optimization of the approach was conducted via a joint loss function that combined cross-entropy loss and a supervised contrastive loss.

Main Results: The quantitative and qualitative analysis of the proposed and other state-of-the-art techniques like SVM, SSRN, DFSL, S3Net, HTLN, DCLN, and CPPM are validated on four benchmark hyperspectral image (HSI) datasets known as Pavia University (PU) dataset, Salinas (SA) dataset, Indian Pines (IP) dataset, and LongKou (LK) dataset. Experimental results indicate that the proposed model outperformes other techniques in terms of classification accuracy, boundary identification, and generalization ability, using minimal samples, while also having acceptable complexity.

Conclusions: This work successfully addresses its goal of designing a strong few-shot classification framework for hyperspectral images (HSIs) leveraging deep transfer learning and contrastive learning methods. The proposed model effectively boosts feature learning by leveraging the strengths of ImageNet-pretrained spatial networks, spectral residual learning, and spectral data augmentation methods. Additionally, the Spatial Attention Module (SAM) helps better capture the context at multiple scales, especially in boundary regions. Finally, the hybrid loss function further promotes inter-class distinction and decreases intra-class variations. Experiments conducted on four popular HSI datasets validate the superior accuracy and generalization performance of the proposed approach in few-shot learning scenarios. Even though the approach is effective, there are some issues related to computational complexity and adaptability for low-resolution and cross-sensor learning that show promising research directions.

Practical Application: The findings of this study can be effectively used in real-world applications of hyperspectral image analysis when there is a lack of annotated data, including precision farming, environmental studies, urban planning, mineral exploration, and disaster response. The developed few-shot learning framework can be effectively used for land cover mapping with only a few training samples, thereby reducing the cost of manual annotation of data. The framework's noise-robust properties make it useful for real-world applications, including UAV- or satellite-based remote sensing systems.

Construction of ASEAN Chess Endgame Tablebases: A Computational Analysis of King-Knight-Queen and King-Bishop-Queen Patterns Against a Lone King

2026-02-17T18:24:16+07:00

Background and Objectives: ASEAN Chess, is a traditional strategic board game that differs substantially from International Chess in terms of piece movement, endgame dynamics, and regulatory constraints. One of its most distinctive features is the official piece-counting rule, which limits the number of moves allowed to force a checkmate when one side is reduced to a lone King. This rule has a significant impact on endgame outcomes, often determining whether a theoretically winning position results in a practical draw. Despite its widespread practice at both national and international competitive levels, ASEAN Chess remains significantly underexplored from computational and game-theoretic perspectives, particularly with respect to complex endgame scenarios involving multiple attacking pieces. Existing studies and practical knowledge rely heavily on human intuition and empirical experience, which are insufficient to fully capture the exhaustive state space of such endgames. The objective of this research is to develop Endgame Tablebases (EGTBs) for ASEAN Chess and to conduct a comprehensive computational analysis of two complex endgame configurations: King-Knight-Queen versus King (K-N-Q vs. K) and King-Bishop-Queen versus King (K-B-Q vs. K). The study aims to determine, under optimal play, the exact distribution of forced wins and draws, to identify the maximum Distance to Mate (DTM), and to evaluate the practical impact of the official piece-counting rules on theoretical endgame outcomes.

Methodology: This research employs a computational framework based on retrograde analysis to exhaustively evaluate all legal endgame positions for the selected configurations. The process begins by generating all legally reachable board states in accordance with the official rules of ASEAN Chess. To manage the large state space, symmetry-based reduction techniques are applied, allowing equivalent board positions to be grouped and analyzed efficiently. Terminal checkmate positions are initialized as Distance to Mate (DTM) zero. The algorithm then iteratively propagates results backward through the game tree, assigning game-theoretic values to each position by considering all possible legal moves for both the attacking and defending sides. Positions are classified as forced wins, draws, or losses under the assumption of perfect play, with the attacking side always moving first. The resulting EGTBs record not only the outcome of each position but also the maximum number of moves required to force a checkmate. In addition, the official ASEAN Chess piece-counting rules are incorporated into the analysis, with particular attention given to the regulation limiting the Bishop’s pursuit to a maximum of 44 moves. This allows a direct comparison between theoretical outcomes and regulation-constrained practical results.

Main Results: The computational analysis reveals that the K-N-Q configuration comprises a total of 11,930,016 legal positions. Of these, only 472,900 positions (3.96%) result in forced wins for the attacking side, while the remaining 96.04% are theoretical draws, indicating strong defensive resilience against a lone King. The deepest winning positions in this configuration require up to 36 moves to achieve checkmate. In contrast, the K-B-Q configuration encompasses 12,170,304 legal positions. Among these, 10,438,976 positions (85.77%) are identified as forced wins, while 14.23% result in draws. The maximum Distance to Mate in this configuration reaches 57 moves, demonstrating that although the position is highly favorable for the attacker, precise and extended maneuvering is often required. When the official 44-move piece-counting rule for the Bishop is applied, 278,040 theoretically winning positions exceed the permitted move limit. As a result, the effective winning probability is reduced to 85.39%, with these positions converted into practical draws due solely to regulatory constraints.

Conclusions: This study provides the first exhaustive Endgame Tablebases for complex ASEAN Chess endgames involving Knight-Queen and Bishop-Queen configurations. The results confirm a substantial disparity in piece potency, with the Bishop-Queen combination exhibiting significantly greater offensive effectiveness than the Knight-Queen combination. Furthermore, the findings demonstrate that official counting rules play a decisive role in shaping practical outcomes, creating a measurable divergence between theoretical optimal play and regulation-constrained competition.

Practical Application: The developed EGTBs establish a rigorous foundation for multiple practical applications. They can be integrated into high-performance ASEAN Chess engines to improve endgame accuracy, utilized as authoritative references for training and instructional systems, and incorporated into digital or mobile learning platforms to enhance players’ understanding of optimal endgame play and counting rules. Additionally, the results provide valuable insights for tournament organizers and rule designers by quantifying the practical effects of regulatory constraints. More broadly, this research contributes to computational game analysis and supports future studies of more complex ASEAN Chess endgames involving additional pieces.

Development of an Anti-Bedsore Bed with Alternating Pressure Points and a Sub-Bed PID-Controlled Cooling System

2026-04-16T14:35:15+07:00

Background and Objectives: Bedsores are injuries caused by the death of skin cells and underlying tissues due to prolonged pressure on specific parts of the body, which reduces blood circulation to those areas and results in insufficient oxygen supply to the tissues. Additionally, moisture combined with long-term pressure can cause skin maceration, making it more susceptible to developing bedsores. If the condition becomes severe and bacteria enter the wound, causing an infection, it becomes difficult to treat and poses a risk of death from septicemia. Therefore, the present research aimed to design and develop an anti-bedsore bed, as well as to evaluate its efficiency.

Methodology: This research was divided into two phases. The first involved designing an Anti-Bedsore Bed based on the principle of alternating weight-bearing points to distribute pressure, which is the primary cause of bedsores, while the second phase focused on performance testing. The evaluation included a load-bearing test where test objects were placed on the mattress base with weight increments of 50 kg up to 150 kg, maintaining each load for 8 hours to study the deflection of the mattress base. Additionally, an under-bed cooling system was tested at a simulated ambient temperature of 33 °C to identify the optimal parameters ( and ) for achieving a rapid temperature response toward a target temperature of 28 °C. Finally, user perception was assessed by having 40 healthy volunteers, categorized by gender and BMI, lying on the Anti-Bedsore Bed for 30 minutes during the operation; data were collected through structured interviews covering their awareness of the bed’s alternating movement, vibrations during operation, physical discomfort, and heat sensation at the skin-mattress contact area.

Main Results: The Anti-Bedsore Bed is divided into three primary sections, namely, a stationary outer frame composing of pillars and beams that support the inner structure and utilize limit switches to regulate movement of the inner bed; a mobile inner bed featuring two sets of interleaved mattress bases that move vertically in alternating loops to distribute the patient’s body weight at set intervals; a cooling unit installed beneath the bed that employs a PID (closed-loop) controller system to adjust fan speed for optimal patient comfort and sweat reduction. Performance testing demonstrated that the mattress base maintained a deflection of 0.236 mm under a maximum load test of 150 kg, while the PID-controlled cooling system achieved its best performance with parameters and set to 90, 15, and 0.1, respectively, resulting in a response slope of -0.101 ºC/s and a steady-state error of only 0.25 ºC. Furthermore, subjective evaluations from healthy volunteers indicate that the physical sensation or discomfort experienced while lying on the shifting mattress bases remained at a low level.

Conclusions: The Anti-Bedsore Bed is composed of three primary elements, i.e., a fixed outer frame, a movable inner frame, and a specialized cooling system, all of which function by alternating pressure points through two sets of mattress bases that interchangeably shift upward and downward. This equipment can support a maximum load of 150 kg and is engineered to rapidly lower and stabilize the ambient air temperature beneath the bed without significant fluctuations. Clinical trials involving volunteers indicate that while both male and female users perceived the mechanical movements at a similarly low level, individuals with Body Mass Index (BMI) over 30 reported more extensive heat and discomfort compared to other groups, a result attributed to their physiological tendency to accumulate more body heat.

Practical Application: An Anti-Bedsore Bed prototype has been developed to benefit bedridden patients by effectively reducing the occurrence of bedsores.