In Addis Ababa, Ethiopia, pedestrians account for more than 80 percent of all fatal crashes. Speed humps have been implemented citywide since the launch of a road safety strategy in 2017 to calm vehicular traffic and improve road safety. The city targeted major roads and installed over 550 round-top asphalt concrete speed humps with a height of 10 cm and lengths greater than or equal to 4.9 m. This study assessed the impact of this citywide intervention in terms of crash reduction. A before-after design was used, analysing geospatial data on speed humps and fatal crashes for three years before and after implementation. A 250 m buffer is created around the speed humps using QGIS to represent their influence areas. The Join attributes by location (summary) tool was used to count the total number of crashes within these buffer areas. Fatal crashes decreased by 26 percent within the speed hump influence areas, while they increased by 13 percent outside the areas. The installation of speed humps was estimated to have prevented 135 fatal crashes over three years, corresponding to approximately 142 lives saved. Further analysis assessed the relative change and its significance showing a 35 percent reduction in fatal crashes in speed hump influence areas compared to outside the areas using the Tanner test (k = 0.65). The chi-squared test indicated that the difference was statistically significant (at a 95% confidence level). Given the limited availability of alternative interventions, it is reasonable for cities in low- and middle-income countries to continue installing speed humps alongside complementary measures such as traffic signals and speed management technologies. This aligns with the principle that human life should not be compromised for mobility.
Horizontal curves are a significant contributor to crash risk on rural road networks. Historically, high-risk curves are identified reactively, based on examination of crash history. However, to reduce the influence of random variation in observed crashes, proactive approaches, which estimate underlying crash risk, are increasingly preferred. To efficiently target road safety investment, a scalable, proactive approach to estimating rural crash risk is required. This paper presents a proactive corridor prioritisation framework for rural roads in the United States of America (US), based on horizontal curve risk. Road centreline geometry and speed limit data were used to estimate typical operating speeds and “curve context” – the degree to which the estimated operating speed of a curve differs from its safe traversal speed. A Safety Performance Function was used to estimate underlying risk, based on curve context and other geometric and traffic variables. Estimation and validation of the model using Arkansas and California datasets indicated that poor curve context is a significant predictor of crash risk. A case study of the Arkansas network is presented. The model allows estimation of crash risk across the US without reliance on crash data. Corridors are prioritised according to predicted out-of-context curve crash risk per unit length. This approach highlights corridors where treatments to address curve context are expected to deliver the greatest crash reductions relative to treatment costs. A map-based interactive tool is used to display corridor priorities. Targeting safety improvements on prioritised corridors is expected to produce significant reductions in rural road crashes.
The safety of vulnerable road users (VRUs), including pedestrians and cyclists, remains a persistent challenge in complex urban traffic environments. While vehicle-to-everything (V2X) technologies have improved safety for vehicle occupants, VRUs are often not part of these systems and have not benefited from the technology-based safety gains. The widespread use of smartphones presents an accessible platform for delivering real-time safety alerts. This study investigated user perceptions, acceptance, and privacy attitudes toward smartphone-based vehicle-to-pedestrian (V2P) communication systems. A quantitative survey was conducted among road users (N = 153) with experience in Australia and China, capturing perspectives across different traffic environments and technology contexts. Semi-structured expert interviews were also conducted with participants holding academic or industry expertise (N = 5) to provide contextual insights into user expectations, interface design, and deployment considerations. Results show limited prior awareness of V2P applications but strong interest in adopting smartphone-based alerts for pedestrians, provided that systems are simple, responsive, and minimally distracting. High levels of acceptance were observed across participants, including those without prior experience of V2P alerts, suggesting that familiarity is not a prerequisite for adoption. The importance of trust and perceived control was highlighted as attitudes about privacy were distributed across unconditional acceptance, conditional acceptance, and refusal of location sharing. The study identifies key human-centred design priorities for smartphone-based safety systems and supports their feasibility when designed around usability, reliability, and privacy, addressing a gap in existing research that has predominantly focused on technical system performance. Findings provide practical insights for developing inclusive, scalable, and privacy-aware safety solutions for VRUs within intelligent transportation systems (ITS).
This study analysed police-reported bus crash data in Dhaka City, Bangladesh from 2015 to 2022, highlighting the significant class imbalance between fatal and non-fatal incidents that challenges traditional machine learning approaches. To address this gap, this study evaluated six ensemble frameworks combining tree-based models (Random Forest, Gradient Boosting, and XGBoost) with baseline classifiers (Logistic Regression, Support Vector Machine, and CART). Various sampling techniques were used, including non-sampling, resampling (SMOTE, Random Oversampling), and training strategies (No Bagging, Bagging, optimised via GridSearchCV). Results show that Logistic Regression, when optimised with GridSearchCV and combined with SMOTE resampling, achieved the best performance metrics: accuracy = 0.745, AUC = 0.697, F1-Score for Non-Fatal = 0.840, and Precision for Fatal = 0.600. This model outperformed ensemble methods, such as Gradient Boosting (AUC = 0.702) and Random Forest (AUC = 0.680). While Gradient Boosting effectively handled feature interactions, Logistic Regression demonstrated a 13% improvement in accuracy for fatal cases with hyperparameter tuning. Resampling had a limited impact on tree-based models; for example, the AUC of Random Forest increased marginally from 0.643 to 0.648 with bagging. Moreover, the Support Vector Machine struggled with precision in fatal cases, with values ranging from 0.000 to 0.420, even with random oversampling. The key predictors identified included pedestrian casualties, vehicle fitness certification, driver age, seatbelt use, drunk driving incidents, and road geometry. These findings support a model for predicting bus crash injury severity in Dhaka City and similar urban areas, emphasising the need for data-driven strategies to prevent crashes that result in fatality or serious injuries.
Children with disabilities and medical conditions are at an increased risk of injury or death during road vehicle transportation. In 2020, a national Australian survey of health professionals and organisations involved in the safe transport of children with disabilities and medical conditions identified a critical need for improved access to training, education, and resources. In response, Mobility and Accessibility for Children and Adults (MACA) Ltd developed targeted information and training resources. This study examined the impact of these resources on current road safety practices in Australia. A follow-up national survey was conducted with health professionals (n=126) and organisations (n=40) involved in supporting the safe transport of children with disabilities and medical conditions. Respondents included 74 participants (health professionals, n=56 health organisations, n=18) who also completed the 2020 survey, enabling direct comparison over time. Statistical and descriptive analyses of repeat participants reported greater confidence in their knowledge and more frequent assessment of transport needs for children with disabilities and medical conditions. The results reported a reduced frequency of barriers like lack of specialised knowledge, difficulty gaining funding for equipment, insufficient training and professional support, and unclear processes. However, many challenges persisted, with most respondents still reporting barriers. Health professionals and organisations that accessed the resources generally found them helpful, suggesting their potential benefits. Overall, the findings indicate promising improvements in the safe transport for children with disabilities and medical conditions, but additional efforts are required.
