Engineering

Abstract: Eco-driving at signalized intersections uses signal timing to plan vehicle arrivals, smooth speed changes, and reduce CO₂ emissions. However, the effectiveness of eco-driving policies may be limited without considering queue discharge, because queues ahead can delay vehicle passage even during a green phase. This study proposes a queue-aware multi-agent eco-driving model for connected and automated vehicles (CAVs) in mixed traffic. First, an effective green window is estimated for each CAV by accounting for queue discharge. Second, a local-emission reward combines emissions from the controlled CAV and nearby same-lane followers. Finally, these components are integrated into a proximal policy optimization controller that generates continuous acceleration commands. The proposed model is evaluated in SUMO across varying traffic demands and CAV penetration rates. Results show that CO₂ emissions reduce in most tested conditions and the reduction rates grow with CAV penetration at each demand level. At 1200 veh/h, the reduction rises from 7.0% to 25.6% as CAV penetration increases from 10% to 70%. Ablation results further reveal that the effective green window and local-emission reward jointly contribute to the CO₂ reductions. These findings highlight the potential of the proposed method for eco-driving in mixed traffic.

Abstract: This study presents an axle-based, uncertainty-aware safety-envelope framework for the Automatic Braking and Stopping System (ABSS), designed to arrest rolling stock before a hazard boundary after uncontrolled movement begins. The method combines free-rolling dynamics, response delay, active braking, deterministic force sizing, axle-capacity screening, and uncertainty-adjusted distance assessment. It separates gross mass, braking demand, and verified eligible axle positions. A browser-based tool performs these calculations and configuration checks. A predefined distance serves numerical verification; operational use requires the measured detection-to-boundary distance. For illustrative 94 t/4-axle, 188 t/8-axle, and 282 t/12-axle cases, the profile yields a detection speed of 7.973 km/h. With four 10-kN elements, deterministic stopping distances are 10.211, 16.553, and 23.277 m; uncertainty-adjusted upper distances are 10.698, 17.430, and 24.636 m. Deterministic sizing requires two elements in each case, while uncertainty-adjusted sizing requires three for 282 t. Restricting that case to two eligible positions raises the upper distance to 52.417 m and leaves no feasible configuration. Sensitivity is greatest to detection speed, followed by braking force and gross mass. Thus, distance adequacy alone is insufficient: uncertainty and physical installation capacity must also be checked. The framework supports traceable screening, software verification, and subsequent component, integration, and controlled-track validation.

Abstract: This paper presents the design, implementation, and preliminary assessment of a three-space Internet of Things (IoT) smart parking prototype motivated by parking-management needs in Amman, Jordan. The system combines an ESP32 development board, infrared (IR) occupancy sensors, an RC522 radio-frequency identification (RFID) reader, light-emitting diode indicators, a liquid-crystal display, servo-operated model barriers, and a Blynk interface. IR sensing provides space-status information, while three colour-coded RFID tags are associated with spaces P01–P03 to support a parking-location reminder. The documented demonstrations illustrate parking-state changes and the display of tag-associated locations. An exploratory demonstration involving ten users reported 80% satisfaction, suggesting interest in the concept without establishing population-level acceptance. Preliminary technical indicators are presented with their reporting limitations because trial counts and event-level records do not accompany the study. Larger-facility search-time comparisons are retained separately as illustrative scenarios rather than measured outcomes. The contribution is an integrated proof of concept that distinguishes occupancy detection, tag identification, and user information within a small parking model. The implementation provides a basis for further development, but reproducible bench testing, an ESP32-specific wiring record, stronger identity-to-space association, and controlled field evaluation are needed to establish operational reliability and actual parking-search-time benefits.

Abstract: Charging infrastructure for battery electric bus (BEB) fleets must be sized against the timing of demand, not only its magnitude: an accurate per-trip energy estimate cannot on its own establish whether a depot’s chargers, contracted grid connection, and dispatch schedule can recover the fleet’s energy deficit without breaching demand-based tariff thresholds. This is especially consequential in resource-constrained settings, where notified-maximum-demand penalties and steep time-of-use differentials turn a mistimed charging load into a direct commercial liability rather than an operational inconvenience. Yet pre-investment planning tools typically treat vehicle energy consumption, dispatch scheduling, and charging infrastructure as separate problems, so a validated demand estimate rarely translates into a validated depot load profile. This paper presents an operational digital twin that converts the per-trip arrival state-of-charge predicted by a companion rolling stock twin into a depot-level charging load profile for the Golden Arrow Bus Services (GABS) electric fleet at the Arrowgate depot in Cape Town, South Africa. The twin reconstructs each bus’s daily duty chain from scheduling data to determine its available charging window, then simulates session initiation, power draw, and duration against these windows using charging behaviour parametrised from depot Open Charge Point Protocol (OCPP) records, including the smart charging constraints already enforced at the depot. Each bus’s state of charge is carried from one depot visit to the next, so an undercharged departure reappears as a lower arrival rather than disappearing from the model. The simulated load profile is validated against 968 OCPP session records from two periods with different depot hardware, achieving energy-per-session agreement of 4.3 to 7.3 % and load-shape correlation of 0.95 to 0.99, and is evaluated against the depot’s contracted notified maximum demand of 3 000 kVA. Scenario analysis quantifies how the charging schedule, charger rate, and site pool limit shift the load profile relative to this threshold and to time-of-use cost exposure, showing that scheduling charging into off-peak windows without a site-wide limit can raise peak demand above uncontrolled charging and beyond the contracted threshold (3 167 against 1 832 kW), that a site pool limit controls peak demand and monthly cost together, and that a 30 kW per-charger cap leaves buses unable to complete up to 15 % of their trips. By coupling a validated, condition-aware demand layer to depot dispatch and charging logic, this work supports capital-efficient charging infrastructure decisions for transit operators in the Global South, where undersized infrastructure risks service disruption and oversized infrastructure diverts capital from wider electrification.

Abstract: Urban traffic congestion in the vicinity of educational institutions is largely determined by the use of cars for school trips. The paper analyzes the impact of the implementation of the Pedibus program on the performance of an urban intersection in the municipality of Pitești, using real traffic data and microscopic modeling in the PTV Vissim environment. The results indicate that the intersection currently operates at service level F, with high delays and capacity overrun. The implementation of the Pedibus program, modelled through a reduction in school-related vehicle flows based on the school mobility survey and the assumed participation scenario, leads to improvements in intersection performance and a transition from LOS F to LOS E. The study highlights the effectiveness of mobility measures based on changing user behavior, demonstrating their potential to reduce congestion in school areas without major interventions on the infrastructure.

Abstract: National freight frameworks consider volume, economic value, and network connectivity, but public data do not directly identify freight dependence at individual railroad bridges. This gap limits consistent national screening for preservation and resilience planning. This study developed a reproducible national workflow to estimate bridge-level freight exposure and identify regional concentrations of important bridges. The workflow assigned Freight Analysis Framework interzonal rail flows to a connected rail network in the contiguous United States, associated eligible bridges with routed edges, and calculated tonnage, value, and distinct origin–destination (OD) reach. These measures formed a zero-preserving, percentile-based rail-bridge freight infrastructure importance index (RB-FII) that gave equal total weight to economic scale and network reach. Tie-inclusive selection and density-based clustering identified important bridge groups, while parallel-track and component-weight sensitivity analyses tested stability. The model routed all 4,911 retained OD pairs, representing 846.80 million short tons and $370.29 billion, and linked 68,326 of 69,356 processed bridges (98.52%) to the routable network. The nominal top 1,000 cutoff yielded 1,003 bridges after retaining ties; 939 formed 32 regional clusters, and 64 remained regionally isolated. Parallel-track alternatives produced bridge-rank correlations of 0.977–0.979 and identical cluster relationships among common bridges. High-emphasis index component weight scenarios retained correlations of 0.988–0.995 with the baseline. The workflow provides an auditable national screening method for directing detailed engineering, operational, and resilience assessments toward bridges and corridors with high modeled freight dependence. The RB-FII is conditional on modeled routes and does not measure observed train routing, bridge condition, failure probability, or disruption loss.

Abstract: Vehicle dispatching is essential for maintaining supply–demand balance in ride-hailing systems, but existing frameworks separate demand prediction from vehicle allocation, causing a mismatch between prediction accuracy and operational performance. This study proposes a decision-focused end-to-end learning framework that integrates spatiotemporal demand representation and vehicle-to-region dispatching within a unified trainable architecture. The model combines a multi-graph spatiotemporal prediction module with a recurrent Transformer-based dispatching module that iteratively refines vehicle assignments under capacity-aware constraints. A structured loss jointly considers demand prediction, dispatch objectives, and constraint violations, reducing dependence on expensive optimization labels. Experiments on real ride-hailing datasets from Chengdu, Xi’an, and Haikou show competitive dispatching quality and zero capacity-constraint violation rate (CCVR) in the reported city-level comparison. Relative to the Kuhn–Munkres benchmark, the proposed model reduces average online computation time from 242.47 s to 2.47 s, a 98.98% reduction (approximately 98.2-fold acceleration). The decision-oriented mechanism yields operationally useful demand representations, suggesting neural dispatchers offer a scalable approximation for high-frequency operations where responsiveness and feasibility are critical.

Abstract: Conventional multimodal driver-workload recognition relies on behavioral and physiological statistics, whereas discrete task labels do not explicitly encode risk, stage, or strategy. We propose ACT-R Cognitive Semantic Lightweight Fusion (ACTR-CSLF), which aggregates dynamic states from an ACT-R driving model into 20 task-cognitive semantic features and combines them with 16 driving, eye-tracking, and electroencephalographic features in multinomial logistic regression. Leave-one-subject-out validation used 420 weather-task segments from 30 participants, with condition-prior, one-hot, random-code, rank-matched, shuffled-semantic, and leave-one-scene-out controls. Macro-F1 was 0.5795 for Human-only, 0.7485 for ACT-R-only, and 0.7277 for Human+ACT-R; Condition-only reached 0.7674, revealing a strong closed-set task prior. Standard random codes averaged 0.7309, so superiority to arbitrary high-dimensional task codes was unsupported. However, the real ACT-R representation exceeded the empirical 95% upper bound of rank-matched random codes (0.7277 vs. 0.7251; one-sided empirical p = 0.0198), and all 30 shuffled mappings performed worse. In leave-one-scene-out validation, Human+ACT-R outperformed discrete unknown fallbacks in Macro-F1 (0.2590 vs. 0.1902-0.1929) and accuracy (0.5643 vs. 0.2690-0.2714). With 111 parameters, 1.0879 KB storage, and 0.2147 ms mean CPU latency, ACTR-CSLF offers a lightweight, traceable balance of predictive performance, cognitive interpretability, and deployment cost.

Abstract: This article explores the impact of weather and environment on shared bicycles. Using a random forest model combined with explanatory machine learning methods, the relationship, threshold effect, and interaction effect between weather factors and the transfer volume of shared bicycles at subway stations are analyzed; Research has shown that using the RF+IML method to study the impact of weather variables on shared bicycle transfer volume is feasible. There is a significant non-linear relationship between various weather factors and shared bicycle transfers. Temperature, humidity, and rainfall have specific activation and threshold effects on the number of shared bicycle transfers. When humidity is below 60%, the variation in transfer volume remains relatively stable; however, once it exceeds 60%, the transfer volume drops sharply. When the temperature exceeds 17 °C, its impact tends to reach saturation. Similarly, when rainfall reaches around 20 mm, its adverse effect also approaches the threshold. Temperature is the most important factor affecting the prediction of shared bicycle transfer volume, with temperature, cold weather, and cold forecasts contributing over 35% to the total effect. The interaction effect between temperature and other weather factors accounts for 22% of the total effect.During the long-term service of highway tunnel lighting systems, concurrent failures of multiple luminaires lead to severe deterioration of road surface illuminance distribution. Meanwhile, conventional periodic maintenance schemes suffer from significant time latency, failing to ensure driving safety during the maintenance latency period. To address this challenge, this study proposes a highway tunnel lighting quality optimization method based on luminous flux redistribution. First, based on the spatial geometric relationships of failed luminaires, multiple luminaire failure (MLF) scenarios are categorized into four typical topological structures: unilateral continuous failure (UCF), unilateral alternate failure (UAF), bilateral symmetric failure (BSF), and bilateral staggered failure (BStF). The degradation characteristics of average road surface illuminance and illuminance uniformity under each topology are quantitatively investigated. Second, with the optimization objectives of maximizing overall illuminance uniformity (U₀) and centerline longitudinal illuminance uniformity (U₁), a multi-dimensional optimization model is established. This model incorporates foundational constraints, including driving safety, physical dimming boundaries of luminaires, and road surface illuminance stability. Furthermore, to mitigate the transverse dark bands, lateral illuminance imbalance, and longitudinal fluctuations induced by MLF, four enhanced constraints are introduced: global minimum average illuminance, macroscopic regional symmetry, cross-sectional minimum illuminance, and longitudinal illuminance gradient limits. To efficiently solve this high-dimensional nonlinear optimization problem, a hybrid heuristic particle swarm optimization-differential evolution-simulated annealing (PSO-DE-SA) algorithm integrating PSO, DE, and SA mechanisms is developed. Finally, systematic experimental verifications are conducted across the four MLF topologies in a unidirectional three-lane tunnel scenario. The results demonstrate that for discrete failure scenarios supported by adjacent operational luminaires, the proposed method effectively mobilizes redundant luminous flux to achieve substantial restoration of lighting quality. Specifically, under UAF, the average road surface illuminance is restored to 153.46 lx, approaching the ideal non-failure state; under BStF, U₁ increases by 25.02%; under BSF, U₁ increases by 23.96%, effectively repairing transverse illuminance faults. However, under UCF involving three or more luminaires, the lighting system reaches its physical dimming compensation limit: the average road surface illuminance cannot be restored to the safety threshold, ensuring lighting quality only on the lane side with operational luminaires. This finding provides a quantitative criterion for determining the triggering threshold of preventive maintenance and luminaire replacement in tunnel operation and maintenance.

Abstract: Shared micromobility can contribute to more sustainable urban mobility by improving access to urban opportunities without relying exclusively on private motorized transport, but the opportunities that can potentially be reached through the transport network may not coincide spatially with the availability of shared vehicles. This study proposes a data-driven microscopic accessibility framework to assess this relationship at the building level. Origin–destination Global Navigation Satellite System (GNSS) data from free-floating shared micromobility services are map-matched and routed over detailed urban networks to reconstruct plausible trips and derive mode-specific travel speeds. Potential Accessibility is then calculated for each building as the opportunities reachable within a 15-min travel-time threshold. A complementary Vehicle Supply indicator is derived from observed trip-end locations reachable within a 3-min walking-time threshold, and the two components are combined into Effective Accessibility. Their spatial correspondence is further assessed using Spearman rank correlation, high-value coverage, and a normalized dissimilarity index. The framework is applied to Bologna and Florence, Italy, considering bicycles, e-bikes, and e-scooters where available. The results demonstrate that network-based accessibility and shared-vehicle supply provide complementary information and should not be interpreted independently. The proposed framework enables localized identification of areas where high accessibility potential is insufficiently supported by the observed shared-mobility service and can support the planning of more sustainable shared-micromobility systems, while providing a basis for future dynamic accessibility analyses incorporating temporal dependency and fleet availability.

Abstract: Rapid-transit stations are usually judged as transport nodes, yet the walkable area around them decides whether residents can meet daily needs nearby. This study treats each station area as a service system and measures the supply of five civic dimensions (sports, commerce, health, green space and schools) within the 15-min walking catchments of the 125 operating stations of the Santiago Metro, Chile. Catchments were routed on a directed, slope-sensitive pedestrian network, and census counts were allocated by cadastral residential floor area. A Civic Deficit Index, defined against the network benchmark, was checked against census active-commuting shares, and an output-oriented data envelopment analysis frontier was estimated with bootstrap bias correction. One standard deviation of deficit was associated with a 2.65-percentage-point lower active-commuting share after controls. Seventeen station areas (13.6%) formed the frontier; mean efficiency was 0.623 (0.525 after bias correction) and mean scale efficiency 0.891. The two measures agreed partially (ρ = 0.446), and four station areas fell in the worst decile of both. Rankings were stable to weights, typology, frontier specification, peer removal, walking threshold and elevation error, whereas agreement with an order-m partial frontier was moderate (ρ ≈ 0.40). A priority matrix links station area to land-use instruments.

Abstract: Santiago's Metro network is undergoing its largest expansion since 1975 through three new lines—L7, L8, and L9—announced in two waves (2017 and 2018) and opening on staggered timelines through 2033. This multi-cohort rollout is a setting where conventional two-way fixed-effects (TWFE) designs are known to misestimate treatment effects through negative-weighting bias, while static net-present-value rules ignore the option value of deferring irreversible construction under uncertainty. We address both using a hexagon-semester panel (2010S1–2025S2; N = 66,624) and a hedonic cross-section (N = 1,345,805) built from Chile's SII cadastre and F2890 registry, combining Callaway and Sant'Anna's group-time estimator with a linear-programming HonestDiD sensitivity analysis and a real-options investment-timing model calibrated on the panel's own empirical volatility and drift. We find a baseline pooled hedonic capitalization elasticity of +4.9% with respect to gravity accessibility, cohort-specific dynamics a pooled TWFE obscures, a non-monotonic anticipatory uplift across socioeconomic-disadvantage quintiles, and a capitalized land-value gain 2.6 times below the real-options investment trigger—indicating anticipated capitalization alone does not yet clear the option-adjusted threshold for accelerating construction, without speaking to the corridor's broader social return. The results caution against treating land value capture as a substitute, rather than a complement, to social cost–benefit justification for transit megaprojects.

Abstract: The receiver-side magnetic coupler in an electric vehicle wireless power transfer system must dissipate Litz-wire copper loss, ferrite-core loss, and eddy-current loss in the aluminum shield plate. After 60 min of continuous operation, the bare coil reached a hotspot temperature of 82.49 ℃. This left a 2.51 K margin below the 85 ℃ design criterion. We investigated passive thermal management using thermally conductive, electrically insulating silicone thermal interface materials (TIMs) at the coil–core and core–shield interfaces. Temperature-dependent loss models were coupled iteratively between ANSYS Maxwell and Icepak to assess TIM thermal conductivity, interface location, and coverage ratio. The two-way calculation converged after four coupling iterations, whereas a one-way calculation overestimated the total loss by approximately 5.7%. TIM-B had a measured thermal conductivity of 2.40 W m⁻¹ K⁻¹. At both interfaces and 100% coverage, it reduced the coil hotspot to 64.54 ℃, a decrease of 17.95 K. Either interface alone retained approximately 84% of the dual-interface temperature reduction. Among the tested coverage ratios, 110% produced the lowest hotspot temperature and the highest temperature uniformity index. These results provide quantitative guidance for selecting TIM locations and coverage in passively cooled wireless charging receivers.

Abstract: Freight corridors are widely assumed to be resilient when multiple route alternatives exist between key logistics nodes. However, route multiplicity—the existence of several candidate paths—does not guarantee structural redundancy in the graph-theoretic sense, where independent paths must not share critical intermediate nodes or edges [1,2]. Recent infrastructure disruptions in European logistics networks have exposed this assumption as operationally dangerous, motivating a rigorous structural analysis of apparent corridor alternatives [3,4]. A directed corridor graph was constructed from filtered road-network data covering the Benelux region and northern France. An expanded origin–destination audit was performed using nine anchor cities, generating 72 ordered OD pairs, or 71 after excluding one degenerate pair identified during data validation. For each pair, shortest paths, near-shortest alternatives, overlap metrics, bottleneck frequencies, and exact node connectivity were evaluated. Of the 71 corrected OD pairs, 28 were reachable and 27 exhibited at least two near-shortest alternatives. All retained candidate paths fell within 0.40% of the shortest-path cost, and the alternatives overlapped heavily, repeatedly traversing the same nodes and edges. Despite this multiplicity, exact connectivity validation across all 28 reachable pairs showed node connectivity equal to one for 27 of them, confirming dependence on single structural bottlenecks. The sole exception, Antwerp–Lille (k = 2), proved on closer inspection to be geographically fragile in its own right: its two formally independent paths converge within 60 metres of each other at the Lille motorway gateway. These findings have direct implications for infrastructure resilience auditing: route counting alone is insufficient to assess corridor robustness, and targeted structural investment in identified bottleneck locations is required to achieve meaningful redundancy.

Abstract: Metro systems have become increasingly crucial components of sustainable urban mo-bility solutions. The effectiveness of metro systems, however, depends on station ca-pacity and passenger behaviour within stations is a recurring and constantly evolving challenge. Efficient crowd management in metro stations is therefore a critical chal-lenge for transport operators due to increasing passenger demand and the need to maintain safe and reliable operations. Using a combination of agent-based modelling and discrete-event simulation, this study develops an integrated simulation frame-work, and applies to two typical metro stations, De Ferrari Station in Genoa, Italy, and Serdica Station in Sofia, Bulgaria, using Simul8 – an event-based simulation modelling software. The models represent passenger and train interactions within station envi-ronments, including inbound, outbound and transferring passenger movements, ticket purchasing and validation activities, utilisation of stairways, escalators and elevators and train boarding and alighting processes. Passenger behaviour is modelled through a combination of specific routings, queuing and stochastic activity durations provided by data from metro operators, site observations and local knowledge. Validation was undertaken by comparing simulated passenger flows against observed data at key sta-tion locations, including entrances, platforms, concourses, and transfer areas. Results demonstrate a high level of accuracy, with simulated flows closely mirroring observed values for both stations. Findings confirm that the integrated simulation framework provides a robust and effective approach for analysing station operations, identifying congestion points, and supporting evidence-based decision-making for station design and crowd management improvements.

Abstract: We study the topology-dependent performance of scheduling-aware multipath routing in capacitated networks subject to congestion, node throughput constraints, downstream storage limits, and structural failure. We evaluate four routing strategies: shortest-path routing (SP), k-shortest paths (KSP), vertex-disjoint routing (VD), and a spillback-aware scheduling heuristic (VDSched). Experiments are conducted across five canonical synthetic topologies: grid, Erdős–Rényi random, scale-free, small-world, and spatial random networks, and four stress scenarios: baseline, random edge failure, targeted edge failure, and capacity degradation. The primary suite uses target n = 400-node networks, 50 origin–destination pairs per instance, 20 instances per topology across four graph-generation seeds, yielding 400 seed–instance realisations and 12,800 primary simulation runs. Statistical comparisons use paired Wilcoxon signed-rank tests with Holm correction.VDSched reduces peak blocked mass in every topology–scenario combination tested, with reductions of 36.5–87.0% relative to SP in the primary η = 5 suite. Its effect on mean delivery time, however, is strongly topology dependent. VDSched reduces mean delivery time by 18.9–23.6% in Erdős–Rényi random networks and by 31.6–36.7% in spatial random networks. In scale-free networks, it reduces delivery time by 9.8–10.4% under baseline, random-failure, and capacity-degradation scenarios, but provides no significant benefit under targeted removal of structurally important edges. In small-world networks, delivery-time gains are weaker and scenario-dependent, with the clearest benefit under capacity degradation.Grid networks reveal the opposite regime: VDSched increases mean delivery time by 1.5–3.4% relative to SP despite reducing peak blocked mass by 68.4–77.6%. This exposes a latency–congestion trade-off in topologies where alternative routes are geometrically inefficient. A storage-sensitivity analysis across η ∈ {3, 5, 8} confirms that these topology-dependent performance regimes are robust to variation in node storage capacity, with topology remaining the dominant moderator of delivery-time performance across all storage settings. These findings establish that VDSched is a robust congestion-suppression mechanism whose latency benefit depends on whether the topology provides efficient alternative routes.

Abstract: Micromobility has become an increasingly important part of sustainable urban transport, but its rapid expansion has also raised safety concerns for users who have limited physi-cal protection in the event of a crash. National-level evidence on micromobility safety re-mains limited in Central and Eastern Europe. This study therefore provides an epidemio-logical assessment of 836 micromobility-related crashes recorded in Hungary in 2024 us-ing administrative data from the Hungarian Central Statistical Office (KSH/HCSO). De-scriptive statistics were used to characterize injury severity, hourly crash occurrence, weather conditions, nature of accident, and road shape. Chi-square tests were then ap-plied to examine whether injury severity was associated with these temporal, environ-mental, and crash-context characteristics. For inferential analysis, injury severity was di-chotomized into slight versus serious/fatal injury. Monte Carlo permutation procedures were used to estimate p-values, and Cramér’s V was calculated to quantify the strength of association. Slight injuries accounted for 63.4% of recorded crashes, serious injuries for 35.8%, and fatal outcomes for 0.8%. Crash occurrence was concentrated in the afternoon, with 192 crashes (23.0%) recorded between 15:00 and 17:00 and the highest hourly fre-quency occurring at 16:00 (n = 79). Nature of accident was significantly associated with injury severity (χ²(11) = 26.36, p = 0.0037, Cramér’s V = 0.178), whereas weather condition, road shape, and hour of day were not. These findings identify crash mechanism as an important area of focus for future micromobility safety analysis and prevention.

Abstract: Reliable road-accident analysis requires attention to numerical stability, appropriate exposure treatment, and temporal validation in addition to predictive accuracy. This study presents an integrated numerical and machine-learning analysis of road accidents in Bangladesh using four synchronized datasets covering 2023–2025, with 288 divi-sion-month observations and a 36-month national series. The data include 17,046 acci-dents, 15,994 fatalities, and 20,389 injuries. The analysis combines singular-value and condition-number diagnostics, Normal Equation, QR, SVD and Tikhonov solvers, scaled polynomial approximation, spline and Savitzky–Golay methods, Poisson and NB2 like-lihood models, Kalman/RTS state-space estimation, Fourier analysis, and fifteen statistical and machine-learning models evaluated by leave-one-year-out validation. The fatali-ty-design matrix was full rank with κ(X)=16.73, while scaling reduced the cubic Van-dermonde condition number from 6.21×10⁴ to 7.77. Poisson was preferred to NB2 for conditional fatality analysis, and Fourier analysis identified a dominant 12-month cycle. Ridge regression, Huber regression, and Poisson random forest gave the best predictive results for accidents, fatalities, and injuries, respectively. Barishal recorded the highest annualized fatality rate at 5.44 per 100,000 population. These results show that combining numerical analysis with time-aware predictive modelling improves the robustness and interpretability of road-safety assessment in Bangladesh.

Abstract: Access to home-based smart charging may be unequal because smart capability becomes relevant only after households can obtain domestic charging infrastructure. This study examines where the disparity associated with private off-street parking arises across these two nested access stages. We used England’s 2024 National Travel Survey, the first NTS wave to observe both infrastructure stages within the same vehicle-questionnaire route. Stage 1 included all 478 eligible plug-in vehicle records with complete data; 342 records with domestic access and substantive smart-status data formed Stage 2. Survey-weighted logistic models, marginal standardization, and bootstrap inference clustered at the primary sampling unit were used to estimate and decompose the adjusted joint contrast. Private off-street parking was associated with a 41.1-percentage-point higher probability of reported smartcapable domestic charging access (95% confidence interval 27.5–51.8). Of this contrast, 40.6 points were allocated to domestic chargepoint access and 0.5 points to conditional smart status; the latter was imprecisely estimated and does not establish equivalence. The findings indicate that the observed parking disparity arises mainly before smart capability becomes relevant. Policies that expand smart functionality among existing home chargers may therefore leave the underlying domestic-access barrier unresolved. The NTS observes infrastructure states, not actual smart-charging behavior or electricity-system effects.

Abstract: Mobile robots with omnidirectional locomotion have attracted increasing attention owing to their superior maneuverability in confined environments. Although omnidirectional wheeled robots have been widely investigated, research on omnidirectional tracked robots remains limited. One of the key challenges in the design of such systems is the computationally efficient determination of motion resistance resulting from the rollers-surface interaction. Therefore, this study aims to develop surrogate models for predicting the longitudinal and lateral motion resistance of an omnidirectional tracked roller. A multibody dynamics model of a single roller was developed to simulate the contact interaction under different roller orientation angles, applied masses, coefficients of friction, and bearing friction torque. An experimentally validated model was subsequently employed to multiple regression surrogate models relating the operating and design parameters to the motion resistance forces. The developed surrogate models achieved coefficients of determination exceeding 99%, enabling rapid prediction of both longitudinal and lateral forces while preserving high predictive accuracy. The proposed surrogate models provide a computationally efficient alternative to repeated multibody dynamics simulations and support the design of omnidirectional tracked mobile robots.