TIME
SESSION
LOCATION
8:00 – 9:00
Registration and morning networking
Great Hall Foyer
9:00 – 10:30
Opening:
Professor Anika Gauja, Pro Vice-Chancellor Research, University of Technology Sydney
Plenary 2: Humans, Strategy and High Speed Rail
Keynote speakers:
Beau Miles
John Alexander OAM
Professor Lynette Cheah, University of the Sunshine Coast
UTS Great Hall
10:30 – 10:50
Morning Tea
Great Hall Foyer
11:00 – 12:30
Paper presentation period 3
Paper 36
Speed reduction effectiveness of mid-block raised pedestrian crossings using IoT-based speed data
Room Location: B11.B2.101
View Authors
- Yuting Zhang (University of New South Wales)
- Meead Saberi (University of New South Wales)
View Abstract
| This paper uses Compass IoT connected vehicle speed data to evaluate how mid-block raised pedestrian crossings affect vehicle operating speeds and pedestrian safety. A before-and-after design was applied to ten sites in the Greater Sydney metropolitan area, including three within school zones. Speeds were extracted within a 100 m corridor centred on each crossing and analysed using non-parametric tests, stratified comparisons and speed profiles. Statistically significant reductions occurred at all three measurement points after installation of a raised crossing, with the largest effect at the raised platform. Median average speed and median 85th percentile speed decreased by about 4 km/h and 8 km/h, respectively, showing reductions in both typical and higher-end operating speeds. Spatial analysis confirmed the reduction was concentrated at the platform, with limited persistence across the full corridor. Post-installation average speeds at the platform approached the 20 km/h advisory level, while 85th percentile speeds remained closer to the 30 km/h survivability threshold. Raised crossings achieved safer speeds at the conflict point but did not consistently reduce faster vehicles to the advisory speed, nor sustain low speeds along the full corridor. School zone sites showed smaller additional reductions, reflecting already moderated baseline speeds. Case study comparisons showed that physical vertical deflection slowed vehicles more than visual cues alone. |
Paper 58
Spatial proactive diagnosis of urban autonomous driving instability via traffic simulations
Room Location: B11.B2.101
View Authors
- Hoyoon Lee (Korea Institute of Civil Engineering and Building Technology)
- Changhyeon Cho (Hanyang University)
- Minkyung Kim (Hanyang University)
- Jeonghoon Jeend (Hanyang University)
- Cheol Oh (Hanyang University)
View Abstract
| Urban mixed-traffic corridors contain dense intersections, frequent access manoeuvres, and bus-stop dwells, intensifying vehicle interactions. Differences in driving characteristics between autonomous vehicles (AVs) and human-driven vehicles (HVs) can increase trajectory irregularity and localised driving instability. A simulation-driven spatial framework was developed to identify and rank autonomous driving instability segments in an urban corridor for subsequent field investigation and operational and geometric countermeasures. A microscopic traffic simulation with fixed traffic demand and signal timing generated vehicle trajectories at 0.1 s intervals, enabling controlled comparison between AVs and HVs under identical traffic conditions. Sliding time windows were applied to compute driving instability indicators capturing both longitudinal and lateral variations, and AV–HV sample imbalance was addressed through undersampling. Multivariate analysis of variance quantified between-group effects across the indicators, and effect sizes were normalised as weights to construct a composite driving instability indicator. Time points exceeding the 80th percentile were labelled as events. Kernel density estimation of event locations identified hotspot segments, which were concentrated within intersection influence areas rather than being uniformly distributed along the corridor, indicating increased interaction complexity and potential safety risks. Top-ranked segments shared recurring interaction-amplifying conditions, including bus-stop proximity, overlapping bus routes, protected left-turn and U-turn operations, and multi-lane configurations. The framework supports proactive pre-deployment diagnosis by prioritising autonomous driving instability segments and provides a transferable approach for evaluating mixed-traffic environments. |
Paper 101
Safety perception in multimodal streets using structural equation modelling
Room Location: B11.B2.101
View Authors
- Amir Mortazavi (University of Technology Sydney)
- Niousha Bagheri (University of Technology Sydney)
- Milad Ghasri (University of New South Wales)
- Cecilia da Rocha (University of Technology Sydney)
- Kasun Wijayaratna (University of Technology Sydney)
View Abstract
| Urban street design has increasingly shifted from vehicle-oriented traffic management toward approaches that balance mobility with place-making and pedestrian activity. In this context, shared and multimodal streets have emerged as an alternative design paradigm that promotes interaction between road users within a common environment. While previous research has largely focused on objective safety measures such as crashes and vehicle speeds, perceived safety remains a critical behavioural factor influencing users’ willingness to engage with these environments. This study develops a Structural Equation Modelling (SEM) framework to examine how perceived street characteristics influence safety perception in shared street environments. Using data from a face-to-face intercept survey conducted across 12 case study sites in New South Wales, Australia (n = 641), the model investigates the relationships between Street Design Quality (SDQ), Street Functionality (SF), Perceived Safe Speed (PSS), and Safety Perception (SP). Overall, the findings indicate that safety perception in shared street environments is primarily shaped by physical design characteristics and perceived operating conditions. From a planning perspective, the results highlight the importance of pedestrian-oriented design, environmental quality, and intuitive speed management in creating safer and more comfortable urban streets. |
Paper 117
Factors influencing pedestrian-AV interactions in shared spaces: A synthesis of the literature
Room Location: B11.B2.101
View Authors
- Dhanushka Sandaruwan (Deakin University)
- William Payre (Deakin University)
- Ben Horan (Deakin University)
- Christophe Bastien (Coventry University)
- Ashim Debnath (Deakin University)
View Abstract
| Pedestrian interactions with automated vehicles (AVs) in shared spaces have emerged as a pivotal research domain driven by the rapid deployment of AV technologies and the growing global challenge of pedestrian safety in urban mixed traffic environments. Unlike conventional traffic environments, shared spaces rely heavily on informal negotiation, implicit communication, and mutual interpretation between road users. Consequently, understanding the factors influencing pedestrian-AV interactions in these environments is essential for improving pedestrian safety and operational efficiency. However, the current literature remains fragmented, with a notable lack of consensus regarding the factors influencing pedestrian-AV interactions. To address this gap, this review synthesises the literature on pedestrian-AV interactions in shared spaces, categorising the influencing factors into five thematic categories: pedestrian characteristics, AV characteristics, shared space characteristics, environmental characteristics, and interaction dynamics. The findings show that pedestrian demographics, AV and pedestrian kinematics, and visual eHMI design factors are the most commonly used factors in the literature. Conversely, several important factors remain largely underexplored, including free-form pedestrian trajectories, group negotiation behaviour, and two-way shared-space dynamics. The findings suggest that future research should focus on integrating immersive simulations, empirical trajectory data, and psychological perception measurements into socially aware hybrid modelling frameworks to analyse pedestrian-AV interactions in shared spaces. |
Paper 80
Are cities ready for autonomous buses and shuttles as public transportation? A systematic review of public acceptance and implementation readiness
Room Location: B11.B2.103
View Authors
- Amanda Ristriana Pattisinai (University of Auckland)
- Mohsen Mohammadzadeh (University of Auckland)
- I-Ting Chuang (University of Auckland)
View Abstract
| This study aims to examine the city readiness and public acceptance of Autonomous Buses and Shuttles (ABS) integration within urban public transport systems through an integrative systematic literature review. Using the PRISMA framework, relevant studies were identified, screened, and assessed to synthesise the key factors influencing ABS implementation, governance readiness, and societal acceptance. The publications included in the final assessment have been investigated using co-occurrence keywords, geographical distribution, co-authorship, citation reports, the annual number of publications, and the distribution of articles across different themes. Furthermore, the findings show that most studies are published in journals related to transport policy, travel behaviour, and sustainability. Geographically, the literature is concentrated in Europe and the Asia-Pacific, with publication trends increasing significantly in 2025. The discussion further reveals that ABS acceptance is shaped not only by technical considerations such as safety and technological readiness, but also by socio-cultural transformation, symbolic meanings, individual perceptions, and social influence. The study concludes that successful ABS integration requires a broader and more inclusive approach beyond technological capability alone. |
Page 81
Privacy-preserving personalized federated learning for multimodal transportation demand forecasting
Room Location: B11.B2.103
View Authors
- Bing Song (The Hong Kong University of Science and Technology)
- Sisi Jian (The Hong Kong University of Science and Technology)
View Abstract
| Accurate multimodal transportation demand forecasting is important for intelligent transportation system management, but practical model training is often limited by distributed data ownership, heterogeneous demand patterns, and privacy concerns. Federated learning enables collaborative training without sharing raw mobility data, yet conventional federated methods usually learn a single global model and may perform poorly under non-IID multimodal demand distributions. In addition, shared model updates may still leak sensitive local information. This paper proposes a privacy-preserving personalized federated learning framework for multimodal transportation demand forecasting. The framework uses a fine-grained graph convolution recurrent network to capture local spatiotemporal dependencies, adopts meta-learning-based personalized federated optimization to support client-specific adaptation, and incorporates R\'enyi differential privacy through gradient clipping, Gaussian perturbation, and cumulative privacy accounting. Experiments on NYC Bike, NYC Taxi, and BJ Taxi datasets show that the proposed method outperforms conventional graph recurrent models, standard federated learning baselines, and privacy-preserving federated baselines, achieving a favorable balance between forecasting accuracy and privacy protection. |
Paper 116
A comprehensive framework for arterial progression performance assessment and causal diagnosis using fused trajectory and signal data
Room Location: B11.B2.103
View Authors
- Nainesh Sanjeev Advani (Queensland University of Technology)
- Shubham Sharma (Queensland University of Technology)
- Alexander Paz (Queensland University of Technology)
- Ashish Bhaskar (Queensland University of Technology)
View Abstract
| Arterial signal coordination aims to provide uninterrupted vehicle progression along a corridor. Progression performance is typically assessed sequentially by identifying poorly performing corridors, isolating critical intersection pairs, and diagnosing underlying causes. Traditional methods rely on loop detector data to estimate measures such as percent arrival on green and platoon ratio. These measures are projected from upstream detector actuations using assumed travel conditions and are therefore sensitive to detector placement and unable to fully represent actual vehicle experience. Recent studies have explored the use of trajectory data, which provide detailed representations of vehicle movements. However, their application is often limited by low sample sizes, and existing methods commonly analyse only selected trajectories, such as full-length coordinated movements. As a result, partial and side-street trajectories that contain useful information on progression dynamics are often excluded. Consequently, existing methods can assess progression performance but do not provide a systematic framework for diagnosis, limiting their ability to translate performance assessment into actionable insights. To address this gap, this paper presents a framework for progression performance assessment and diagnosis by integrating trajectory data with high-resolution signal cycle information. The framework systematically utilises available trajectory samples to formulate established corridor- and intersection pair-level performance measures while introducing new trajectory–signal fusion measures and diagnostic parameters for multi-resolution analysis and cause identification from corridor to intersection level. |
Paper 130
Proactive road safety intervention in Australia: Predicting risky driving hotspots from connected vehicle data
Room Location: B11.B2.103
View Authors
- Adriana-Simona Mihaita (University of Technology Sydney)
- Clarence Cheung (University of Technology Sydney)
- Artur Grigorev (University of Technology Sydney)
- Tuo Mao (University of Technology Sydney)
- David Lillo-Trynes (University of Technology Sydney)
View Abstract
| Road safety monitoring has historically been reactive, relying on crash-record analysis after fatalities and injuries have already occurred. Proactive identification of high-risk locations and dangerous driving behaviour before incidents occur is a critical but underexplored challenge. This paper addresses this gap using connected vehicle telemetry data from Greater Sydney, Australia, to detect and forecast near-miss risky driving events at the Local Government Area (LGA) level. Risky driving is quantified through g-force thresholds (hard braking $>0.6g$, harsh cornering $>0.47g$, harsh acceleration $>0.5g$), and spatio-temporal heatmaps are constructed to identify high-risk zones. Eight predictive models are benchmarked across three families: ensemble learning (Random Forests, XGBoost, LightGBM), deep learning (LSTM, N-BEATS), and classical time-series methods (ARIMA, Exponential Smoothing, Prophet). ARIMA achieves the lowest mean absolute error (MAE: 162.21), performing comparably to LSTM (MAE: 163.92) and outperforming all ensemble methods, with N-BEATS reaching an MAE of 180.75. These results demonstrate that parsimonious time-series models are competitive with deep learning approaches when training data volume is limited. The study highlights the potential of IoT-based connected vehicle data to support proactive road safety interventions, with Sydney's inner and western LGAs (CBD, Parramatta, Bankstown) identified as persistent high-risk zones warranting targeted policy action. |
Paper 41
Can shared micromobility close the loop for first/last-mile connectivity? Evidence from a staged field intervention
Room Location: B11.B2.104
View Authors
- Richard Buning (University of Queensland)
- Jonathan Corcoran (University of Queensland)
- Dorina Pojani (University of Queensland)
- Abolghasem Azhdari (University of Queensland)
- Le Diem Quynh Pham (University of Queensland)
- Zhengpeng Zou (University of Queensland)
- Scott Lieske (University of Queensland)
- Thomas Sigler (University of Queensland)
View Abstract
| Shared e-micromobility has been widely positioned as a potential solution to the first/last-mile problem in public transport systems, yet empirical evidence on its effectiveness and adoption dynamics remains limited. This study examines a staged, uncontrolled 'Mobility-as-a-Service (MaaS-lite)' field intervention designed to improve shared e-scooter and e-bike access to a suburban rail station in Brisbane, Australia. Three sequential conditions were observed: baseline operations, dedicated parking hubs with station-focused rebalancing, and the same infrastructure combined with a subsidised $1 flat fare. Two independent data sources were analysed: GPS records for 3,454 trips made by N = 1,141 users and intercept surveys with N = 1,467 station patrons. Daily trip rates rose by 52% between the baseline and parking stages and by a further 16% during the subsidy stage. Trips were concentrated within a 1-4 km band and aligned strongly with weekday commuter periods. However, rising awareness did not translate into broad willingness to adopt shared micromobility. Most respondents remained unlikely to use shared devices, with resistance associated primarily with safety concerns, entrenched routines, identity-based perceptions and the relative attractiveness of walking or driving. Ordered logistic regression identified prior micromobility experience as the strongest predictor of willingness, while car-based station access was associated with the lowest likelihood of adoption. The observed pattern is consistent with operational and pricing interventions supporting use among receptive users, but the absence of a control station means that causation cannot be isolated from seasonality, awareness diffusion or other concurrent changes. Shared e-micromobility is therefore best understood as a targeted first/last-mile mode within multimodal networks rather than a universal substitute for station access. |
Paper 49
Travel to primary school in Wellington City: Effects of home-to-school distance
Room Location: B11.B2.104
View Authors
- Sandra Mandic (Wellington City Council)
- Tessa Madden (Wellington City Council)
View Abstract
| Background: Home-to-school distance, perceived safety, and characteristics of the school route impact children’s active travel to school behaviours. This research compared current and preferred modes of transport to school, parental perceptions of traffic and personal safety and school route characteristics for primary school children by home-to-school distance. Methods: Parents of primary school children (n=488; 50% female) from Wellington City (New Zealand) who completed the Pōneke/Wellington Transport Survey in 2023–2025 and lived within 4 km of their child’s school were included in the analysis. Respondents were categorised into three home-to-school distance categories: <1 km (n=236), 1-2 km (n=163), and >2 km (n=89). Results: As home-to-school distance increased up to 4 km, walking rates decreased (<1 km/1-2 km/>2 km: 62%/22%/4%), cycling rates were mixed (9%/11%/4%) and rates of private vehicle travel (22%/50%/45%), public transport (1%/5%/30%) and scootering (3%/6%/8%) increased (p<0.001). With increasing distance, preference for active transport decreased (76%/46%/34%) and the proportion of children using their preferred modes declined (64%51%/47%) (both p<0.001). As distance increased, parental perceptions of traffic safety for walking to school did not change significantly (67%/57%/54%; p=0.159) but declined for biking (32%/26%/25%; p=0.012) and scootering (51%/42%/33%; p=0.008). Personal safety perceptions declined with increasing distance for all three modes (all p<0.001). With increasing distance, fewer parents reported safe walking (72%/64%/52%; p=0.006) and biking routes (37%/26%/24%; p=0.026) to school. Proportion of parents reporting high traffic volume and dangerous crossings along children’s walking and biking routes or high traffic speed along walking routes did not change with increasing distance. Conclusions: Home-to-school distance must be considered when exploring children’s travel to school and parental perceptions of safety and school route. To support sustainable school travel, future efforts and investments should focus on locating schools in residential areas, ensuring safe walking and biking routes within reasonable distance from schools and encourage active transport. |
Paper 52
Identifying high-risk cycling locations by combining anomaly detection ensemble model and KDE methods
Room Location: B11.B2.104
View Authors
- Vu Le (Deakin University)
- Ali Reza Sattarzadeh (Deakin University)
- Narelle Haworth (Queensland University of Technology)
- Douglas Creighton (Deakin University)
- Ashim Debnath (Deakin University)
View Abstract
| Cyclist safety research is often constrained by the scarcity of labelled crash or near-miss data, which limits the use of supervised machine learning methods. Using large-scale naturalistic cycling data from about 1000 Victorian cyclists over 3.5 years, this study introduces an ensemble of nine commonly used unsupervised machine learning anomaly detection methods to identify high-risk road intersections for cycling in Metropolitan Melbourne and Geelong. By combining the results of the nine methods through majority voting, the method improves robustness and reduces bias in capturing atypical cycling events. The detected anomalies are spatially aggregated using Kernel Density Estimation, with adjusted risk scores incorporated to facilitate the identification of high-risk intersections and corridors. This approach enables the data-driven identification of hazardous locations without relying on extensive labelled datasets, offering actionable insights for infrastructure planning and targeted interventions. Results show that 3.34% of the data points are classified as anomalous cycling behaviour. Comparing the top 100 ranked locations identified by the adjusted risk score (ARS) and KDE methods, 78 of the ARS hotspots exhibit higher risk rates than their KDE counterparts. The results demonstrate the effectiveness of unsupervised ensemble learning in translating naturalistic cycling data into practical road safety knowledge. |
Paper 78
Behaviourally calibrated bikeability and cycling equity: a PLS-SEM bike desert and paradise framework for Chicago
Room Location: B11.B2.104
View Authors
- Shima Mafi (University of Technology Sydney)
- Pooria Choobchian (University of Technology Sydney)
- Jason Prior (University of Technology Sydney)
- Erica McIntyre (University of Technology Sydney)
- Ali Mohamadi (University of Illinois Chicago)
View Abstract
| Bike-sharing system (BSS) ridership remains spatially uneven across urban areas in ways that infrastructure provision alone cannot explain. This study develops a behaviourally grounded framework for measuring bikeability and situating it within a broader system of BSS ridership determinants. Three latent constructs, Potential Demand, Bikeability, and Weather Conditions, are estimated simultaneously using Partial Least Squares Structural Equation Modelling (PLS-SEM) with a Cobb-Douglas structural specification, enabling indicator weights to be derived from observed ridership rather than assumed in advance. The empirical application draws on 3,723,167 Divvy trips across 759 Chicago census tracts in 2024, combined with sociodemographic, built environment, and meteorological data. Bikeability emerges as the dominant determinant of ridership (β = 0.730, elasticity = 0.506), with station availability as its single most influential indicator, followed by transit accessibility and building density. Potential Demand exerts a substantial independent effect (β = 0.654, elasticity = 0.356), while Weather Conditions moderate ridership negatively (β = −0.613). The model explains 75.5% of variation in BSS ridership (R² = 0.755) with strong predictive performance (Q²predict = 0.755). The elasticities sum to 1.105, indicating increasing returns to scale from combined improvements. A bike desert and paradise framework translates these results into a spatial equity diagnostic by computing a supply–demand gap score from the PLS-SEM-derived construct scores. Spatial analysis across four methods reveals a pronounced north–south gradient in cycling equity, with desert conditions concentrated in the northwest and central-west, overlapping with lower-income communities historically underserved by cycling infrastructure. |
Paper 2
The unwritten rules of academia: Insights from leading transport researchers
Room Location: B11.B3.101
View Authors
- Chris De Gruyter (RMIT University)
View Abstract
| Academia is fraught with many unwritten rules. Understanding how to go about various research activities, from generating research ideas to supervising research students, is something that is often only learnt on-the-job and through speaking with more experienced academics. The research underlying this paper aims to provide guidance to research students and early-career researchers on the unwritten rules of academia, drawing on insights from interviews with 20 world-leading transport researchers. Findings are synthesised under a number of key themes, including (1) motivations, (2) research ideas and execution, (3) collaboration, (4) funding, (5) supervision, (6) publication, (7) reviewing and editing, (8) conferences and seminars, (9) communication and research impact, (10) time management, and (11) reflections and the future of transport research. In general, much diversity in the findings was found across interview responses. In some instances, differences could be explained by gender, particularly in relation to the unequal burden with service-related roles faced by some females in academia. In other cases, but more rarely, could differences be explained by discipline, geography or career stage, reflecting increased levels of internationalisation and interdisciplinarity in transport research. A common finding was that behind the unwritten rules of academia are real human beings dealing with the realities and emotional strains of academic life. The insights also suggest that there is no single ‘rule book’ or pathway to success in academia and that flexibility in any chosen approach is needed. |
Paper 11
Estimating resilience benefits: A case study on Sydney’s road network
Room Location: B11.B3.101
View Authors
- Alex Svanberg (Transport for NSW)
- Victor Lam (Transport for NSW)
View Abstract
| The resilience of transport infrastructure to external threats and events is a major focus for investment planning in Australia and around the world. This paper presents an approach and case study for the estimation of a resilience benefit in the context of a major disruption to one of Sydney’s key motorways. While the case study relates to disruption caused by super-peak holiday traffic, the overarching approach is suitable for use regardless of the cause of the initial disruption. The case study finds that the costs associated with network disruptions can be significant, even when the frequency of incidents is relatively low. |
Paper 105
Engineering a win-win: Optimal group buying strategies for food delivery platforms and merchants
Room Location: B11.B3.101
View Authors
- Han Fang (The Hong Kong University of Science and Technology)
- Jinshu Cai (The Hong Kong University of Science and Technology)
- Sisi Jian (The Hong Kong University of Science and Technology)
View Abstract
| Group buying (GB) has emerged as a prominent promotional strategy on on-demand food delivery platforms, driven by the growth of e-commerce. This study investigates the impact of GB mechanisms on market dynamics using a Stackelberg game model. We model the platform as the leader, setting the subsidy rate for GB discount costs with merchants. Merchants, as followers, optimize their GB discount and group size in response to merchant's strategy, considering customers' behavior. Our analysis reveals that GB options can enable merchants to achieve higher profits compared to scenarios without GB. Furthermore, we identify the conditions under which platforms can also benefit from offering GB, creating a win-win situation for both platforms and merchants. These findings provide managerial insights into the key factors influencing GB success, including commission rates, delivery cost-sharing arrangements, and service area characteristics. |
Paper 128
The paradox of fairness: Bandit-based dynamic pricing with endogenous fairness concerns in ride-hailing platforms
Room Location: B11.B3.101
View Authors
- Jinshu Cai (The Hong Kong University of Science and Technology)
- Sisi Jian (The Hong Kong University of Science and Technology)
View Abstract
| Ride-hailing platforms employ dynamic pricing to manage supply-demand imbalances and maximize revenue. However, this practice often results in price discrimination—an outcome that faces growing resistance due to customer fairness concerns. This study addresses this challenge by proposing a dynamic pricing framework that incorporates endogenous customer fairness concerns within a multi-armed bandit (MAB) setting, operating solely on limited bandit feedback. We develop an Upper Confidence Bound (UCB) based algorithm to solve this online learning problem and prove that it achieves a sub-linear theoretical cumulative regret bound of $\tilde{O}(\ln(T))$. Numerical experiments validate the algorithm's performance and yield critical insights into the market dynamics of fairness. Our results reveal a ``majority rules'' pricing principle, where intensifying fairness concerns push prices toward the level preferred by the dominant customer demographic. This dynamic leads to the paradox of excessive fairness: in zones with a high-income majority, a pursuit of absolute price parity incentivizes the platform to set a uniformly high price, creating a lose-lose outcome that diminishes both consumer surplus and platform revenue. |
Paper 28
When demand moves more than weather: Bus-network resilience in Greater Sydney across COVID and flood years
Room Location: B11.B3.103
View Authors
- Haoning Xi (The University of Newcastle)
View Abstract
| Public transport systems in Australian cities face increasing exposure to extreme weather, yet large-scale empirical evidence on weather-service performance relationships outside European and North American contexts remains scarce. This paper merges 897 million stop-level arrival records from the GTFS-Realtime feed for Greater Sydney from June 2020 to June 2022 with Bureau of Meteorology gridded daily rainfall and maximum-temperature observations, producing a 673-day weather-operations panel. Three findings emerge for the network-level daily mean delay indicator. First, demand-regime variation associated with COVID-19 dominates the daily-aggregate signal: he Delta lockdown reduced mean delay by 46–65 seconds, an order of magnitude larger in standardised terms than any rainfall coefficient that this paper estimates, and routine rainfall has a near-zero aggregate association with daily mean delay. Second, multi-day extreme floods are nevertheless visible in the tails of the distribution: the February to March 2022 east-coast floods produced a peak excess delay of about 74 seconds with a season- and regime-matched recovery within two days, and the March 2021 Hawkesbury-Nepean floods produced a comparable peak of 58 seconds with the same two-day recovery. Third, severe-delay hotspots concentrate persistently in western and south-western Sydney corridors, with the rank order of stop-level severe arrivals near-identical between dry and rainy days. This pattern supports a persistence reading: the same corridors are bad in either condition. Policy implications follow directly: priority should be placed on multi-day flood preparedness and structural upgrades to western Sydney corridors, rather than uniform weather-proofing. Route- and stop-level weather coupling, sub-daily exposure dynamics, and farecard-based demand proxies are flagged as priorities for future work. |
Paper 60
Bridging the freight data deficit: A hybrid OD matrix inference framework and survey shelf-life metric
Room Location: B11.B3.103
View Authors
- Wei Sun (Adelaide University)
- Akshay Vij (Adelaide University)
- Nicholas Kaliszewski (Adelaide University)
View Abstract
| Transport authorities globally face a persistent road-freight "data deficit," frequently forced to appraise multi-billion-dollar infrastructure using infrequent, aging legacy surveys and continuous, but origin-destination (OD)-blind, traffic counts. To bridge this gap, this paper proposes a modular, hybrid OD inference framework that fuses an interpretable structural prior which is extracted from a discontinued legacy survey, with contemporary heavy-vehicle traffic monitoring. Operating within a computationally efficient single-level optimisation architecture, the framework introduces a novel hierarchical regularisation strategy. This design acts as an economic safety net for data-sparse environments, stabilising the under-determined inverse problem by anchoring micro-level routing to local land-use patterns while allowing macro-level flows to calibrate to observed traffic realities. Utilizing the island state of Tasmania, Australia as a closed-system pilot, the framework successfully generates a 2024 calibrated ODM. Given the unobservability of full-scope ground truth, the inferred matrix is evaluated via algorithmic sanity checks (near-perfect flow symmetry), macroeconomic plausibility checks (adherence to historical trip production bounds), and partial external benchmarking against bottom-up commodity models. Furthermore, an ablation study demonstrates that the hierarchical regulariser is strictly necessary to prevent catastrophic spatial overfitting in sparse environments. Finally, by conducting a controlled longitudinal experiment across a decade (2015–2024) using a consistent sensor subset, we mathematically operationalize the concept of “survey shelf-life”, detecting a clear, almost monotonic structural decay in the legacy prior. Ultimately, this approach equips transport authorities with an extensible framework for continuous OD estimation, alongside an evidence-based strategic metric to objectively optimize future survey investments. |
Paper 71
Extending route choice models to behaviourally informed cycling routing for agent-based simulation
Room Location: B11.B3.103
View Authors
- Sapan Tiwari (RMIT University)
- Afshin Jafari (RMIT University)
- Steve Pemberton (RMIT University)
- Dominik Ziemke (German Aerospace Center (DLR))
View Abstract
| Cycling network evaluation and agent-based transport simulations often rely on shortest-path routing, implicitly assuming that cyclists minimise travel distance. However, empirical evidence shows that route choice reflects trade-offs between safety, comfort, infrastructure, and terrain. This study develops a behaviourally informed cycling routing framework that integrates route-choice modelling with agent-based simulation. Marginal utilities estimated using a Path Size Logit model are converted into link-level impedance and embedded within MATSim, allowing behavioural preferences to influence network-wide routing while holding demand constant. The framework is evaluated against shortest-path routing and observed routes using identical origin–destination pairs. Results show that behaviourally informed routing better reproduces observed route characteristics, with increased use of low-stress links and cycling infrastructure, and reduced reliance on higher-speed and higher-stress environments. At the network level, routing outcomes shift toward more structured and realistic cycling corridors, while average trip length increases only modestly. These findings demonstrate that shortest-path assumptions can misrepresent cyclist behaviour and network utilisation and highlight the importance of incorporating behavioural evidence into simulation-based transport analysis to improve the realism of cycling network evaluation. |
Paper 92
On the simulation of nested-logit error terms
Room Location: B11.B3.103
View Authors
- Jan Zill (Outer Loop Consulting)
View Abstract
| We show how to draw realizations of nested-logit error terms using an exact transform-based simulator and implement the method in ActivitySim's explicit error term simulation mode. The method combines the representation of nested-logit errors as a function of independent positive-stable realizations with the Chambers-Mallows-Stuck (CMS) transform sampler. We describe the algorithm, discuss implementation details, and validate the resulting simulator against closed-form nested-logit probabilities and comparative statistics across multiple test configurations. |
Paper 75
How equitable is public transport accessibility? A spatial analysis with a case study in Taipei
Room Location: B11.B3.104
View Authors
- Robert B.C. Liu (National Yang Ming Chiao Tung University)
- Barbara T.H. Yen (National Yang Ming Chiao Tung University)
- Corinne Mulley (The University of Sydney)
View Abstract
| Transport planning typically emphasises public transport coverage within specific time periods to ensure general service provision. However, this does not guarantee vertical equity, especially for the vulnerable groups. Moreover, in large-scale multimodal networks, accessibility is often measured by a single index that aggregates the effects of each mode. This paper aims to identify an accessibility index which is simple to understand yet also reveals differences in service levels across modes and cross compare the mode-disaggregated accessibility index with the social characteristics. The index of Public Transport Accessibility Level (PTAL) for the Taipei metropolitan area is used to evaluate the accessibility in Taipei, Taiwan, with public transport modes of metro rail, metro bus, and city bus. Income data at a disaggregate level is used as a proxy to measure equity. The results show that in certain regions of Taipei, there are some obvious unequal differences in accessibility where low income areas have low public transport accessibility and high income areas have high public transport accessibility. The paper provides potential improvement strategies. In areas with low income and low accessibility, two spatial patterns are observed. First, some areas are located in industrial areas or nature reserve areas and have lower demand and thus improving accessibility might not be urgent. However, some areas with low income and low accessibility are located around the main city areas (e.g., peri urban area) with dispersed demand and would benefit from improving accessibility by replacing conventional city bus services with a more flexible mode such as demand responsive transport services. Another potential solution is to reallocate transport resources from areas with high public transport accessibility and high income to areas of low public transport supply. The analysis of this study underpins recommendations to policymakers to make changes to improve equity in public transport accessibility. |
Paper 88
Beyond frequency: A capacity-weighted and reliability-adjusted framework for measuring effective public transport accessibility
Room Location: B11.B3.104
View Authors
- Tristan Clark (TVS Architects)
- Supun Perera (The University of Sydney)
View Abstract
| The Public Transport Accessibility Level (PTAL) metric is gaining traction in Australian planning policy, most recently in Victoria's December 2025 reform of parking provisions, and represents a genuine advance in evidence-based land use decision-making. However, PTAL's reliance on service frequency and walking distance alone, treating all services as equivalent regardless of vehicle capacity or reliability, limits its ability to reflect real-world transport performance. In multi-modal networks like South East Queensland (SEQ), these limitations can materially misrepresent the people-moving capacity available to a location. This paper introduces the Public Enhanced Transport Accessibility Level (PETAL): an evolution of PTAL that builds on its foundations by incorporating vehicle capacity, service reliability, route hierarchy, and bidirectional accessibility. Using open General Transit Feed Specification (GTFS) data and spatial analysis, PETAL measures effective accessibility as capacity per hour, operationalised against supportable bedroom density to produce a metric directly applicable to planning decisions. Applied to SEQ, PETAL provides a more defensible basis for identifying land use mismatches: where zoning underdelivers relative to network capacity, where infrastructure investment is underleveraged, and where bedroom density supportable by the network diverges from what planning rules allow. The flood constraint analysis represents a novel contribution: identifying where high-accessibility land is effectively constrained by flood risk, with direct implications for infrastructure investment sequencing and climate adaptation planning. Better alignment between transit capacity and land use regulation directly supports housing supply and economic productivity. The core methodological contribution - measuring effective accessibility as capacity per hour rather than services per hour- has direct application to housing policy, density controls, and network investment prioritisation. PETAL is offered as a transparent, scalable, and policy- ready evolution of existing PTAL practice, transferable to any Australian or international city with open GTFS data seeking a more rigorous basis for transit-informed land use policy. |
Paper 112
A user-centred feature-preference approach to future electric bus development
Room Location: B11.B3.104
View Authors
- Andres Rodriguez Ruiz (Monash University)
- Robbie Napper (Monash University)
- Selby Coxon (Monash University)
View Abstract
| The success of public transport is key in the decarbonisation of the mobility sector. Despite the unique contribution of route buses to public transport networks (flexibility, better reach, relative low investment compared to other modes) user and non-user perception of the service is the lowest among all urban transport modes. The transition from internal combustion engines to zero emission vehicles presents a significant opportunity to improve the bus riding experience as vehicle design and specification come into focus. By studying the needs and expectations of passengers, it is possible to observe preference for feature design that can contribute to improved user satisfaction. Most of the research in bus-based public transport has been oriented toward the essential performance qualities of the system (frequency, reliability, speed), with a relatively reduced volume of research focused on user expectations beyond functionality. Successful products and services not only perform their stated function well but also produce a positive emotional response from the user. For the project, we adopted the Kano model, a feature-preference method, used in this case to prioritise future development of features that may create positive emotional response from the passenger. A survey (n=129) was carried out with electric and diesel bus passengers, with an additional subset of users recruited at car parking to include public transport expectations of private vehicle drivers and users of other modes of transport. Respondents were questioned about their feelings about the presence or absence of five features, each representative of a desired attribute found in a previous literature review. After applying Berger’s coefficients and plotting results on a traditional Kano chart, we see that where safety and system performance are expected, a higher level of information and comfort are not anticipated by the passenger and would become exciting features to have. Features that are part of the basic level of service ranked the lowest likely because their presence is expected. The consideration of insights revealed in this research in future vehicle design could contribute to a better perception of bus-based public transport systems. |
Paper 133
Exploring the cumulative impact of harassment experience on women’s public transport use
Room Location: B11.B3.104
View Authors
- Rumana Sarker (Monash University)
- Graham Currie (Monash University)
- Subeh Chowdhury (University of Auckland)
View Abstract
| Despite growing commitments to equitable mobility, women experience disproportionately higher rates of harassment on public transport than men, highlighting a critical gap in current transport safety and policy frameworks. While prior research has documented the prevalence of harassment, there is limited understanding of how repeated harassment exposure influences long-term mobility decisions, including reduced transit use. This study addresses this gap by examining the influence of harassment types and its cumulative effects on travel decisions across two life stages: early-life (below 18 years) and adulthood, using a gender-based violence (GBV) framework. Data were drawn from an online survey of 250 women rail users in Melbourne, who self-reported experiencing harassment that reduced their rail use. Compared to non-reduced users, this group reported higher exposure to multiple forms of harassment, particularly catcalling, intimidation, and unwanted touching; nearly 70% reported reduced rail use, with one in four indicating a major decline. Regression result show that verbal harassment is the most pervasive and significantly associated with reduced rail use, particularly when experienced across both life stages. More confrontational experiences, such as coercive sexual pressure exhibit strong effects when occurring in early-life, highlighting the enduring behavioural imprint of harassment on later mobility choices. By moving beyond conventional victimisation frameworks, this study contributes to new evidence on how cumulative harassment experiences translate into behavioural change. The findings underscore the need for gender-sensitive, trauma-informed interventions, improved reporting mechanisms, and early prevention strategies to support safer, more inclusive public transport systems and sustain women’s long-term ridership. |
12:40 – 1:20
Lunch
1:30 – 3:00
Workshops
Looking to a more sustainable future: Utopia or a real possibility?
Andrea Pellegrini (The University of Sydney)
Room: B11.B2.104
From participatory mapping to real-world planning – how to enrich spatial information with unheard contexts
Jussi Sjögren (Tampere University, Finland)
Room: B11.B3.104
Uncovering secrets of successful cross-sector collaborations
Sandra Mandic (AGILE Research)
Room: B11.B1.101
Between the alert and the action: Human trust in AI transport safety
Damien Colabattista (Transport for NSW)
Room: B11.B3.103
High speed rail workshop
Prof. Liton Kamruzzaman (Monash University)
Room: B11.B2.101
Autonomous Vehicle Workshop – Tier V/ Toyota
A/Prof. Simona Mihaita (University of Technology Sydney)
Room: TBA
TfNSW operational tour: Rail Operations Centre: (ROC)
A guided visit to the Rail Operations Centre (ROC), the Sydney Trains 24/7 control facility that manages Sydney's suburban rail network.
Hosted by Transport for NSW.
Departure location to be advised.
TfNSW operational tour: Transport Management Centre: (TMC)
Visit the Transport Management Centre (TMC) and experience a behind the scenes look at how Transport for NSW keeps the state moving every day.
The TMC is the operational heart of the New South Wales transport network
Hosted by Transport for NSW.
Departure location to be advised.
TfNSW operational tour: Central Station Transformation & Clock Tower
This walking tour offers a rare opportunity to experience both the historic and contemporary layers of Australia's busiest transport hub. Discover the stories behind Central Station's transformation, and visit the historic Clock Tower, a Sydney landmark that has kept the city running on time since 192
Hosted by Transport for NSW.
Departure location to be advised.
3:10 – 3:30
Afternoon Tea
3:40 – 5:10
Paper presentation period 4
Paper 103
Comparative analysis of deep learning and physics-informed deep learning for estimating vehicle velocity at intersections
Room Location: B11.B2.101
View Authors
- Md Shohel Parvez (PhD Researcher)
- Afroza Parvin Disa (Bangladesh University of Engineering and Technology (BUET))
- Sara Moridpour (RMIT University)
- Richard Tay (RMIT University)
- Amir Shtayat (Jordan University of Science and Technology)
View Abstract
| Accurate vehicle velocity estimation at intersections is essential for traffic analysis. This study presents a comparative framework based on a Convolutional Neural Network (CNN)-based Deep Learning (DL) model and a Multi-Layer Perceptron (MLP)-based Physics-Informed Deep Learning (PIDL) model, supported by object detection and tracking using YOLOv11 and DeepSORT. The detection model achieved strong performance (precision = 0.956, recall = 0.947, mAP@0.5 = 0.962), ensuring reliable trajectory extraction for velocity estimation. Both models were trained under identical conditions with the same number of epochs to ensure a fair comparison, and both demonstrated similar convergence behaviour while learning velocity patterns from trajectory data. The comparative analysis shows that the DL model captures general motion patterns but produces noisy and unstable velocity estimates, resulting in higher errors (MSE = 0.064516, RMSE = 0.254000). In contrast, the PIDL model incorporates governing motion equations into the learning process, significantly improving prediction accuracy and stability. The PIDL model achieved substantially lower errors (MSE = 0.000875, RMSE = 0.029580) and demonstrated enhanced temporal smoothness, with a reduced Velocity Smoothness Index (VSI = 0.003820) compared to the DL model (VSI = 0.065534). These results highlight that integrating physics-based constraints with data-driven models leads to more accurate, robust, and physically consistent velocity estimation. The proposed PIDL framework offers a reliable solution for modelling vehicle dynamics in complex intersection environments. |
Paper 109
Drivers’ speeding behaviour and safety outcomes in school and roadwork zones: A path analysis approach
Room Location: B11.B2.101
View Authors
- Amrit Ghimire (Deakin University)
- Ross Blackman (Deakin University)
- Narelle Haworth (Queensland University of Technology)
- Ashim Debnath (Deakin University)
View Abstract
| Speeding remains a major contributor to crashes, particularly in high-risk environments such as school zones and roadwork zones. However, existing studies largely examine these contexts in isolation and focus on direct associations, offering a limited understanding of the underlying behavioural mechanisms and how they differ across environments. Utilising data from an online survey of drivers (N = 719), this study addresses this gap by developing two separate path analysis models to examine the direct and indirect relationships among driver characteristics, behavioural factors, speeding behaviour, and safety outcomes (near misses and crashes) in school and roadwork zones. The findings revealed a shared safety pathway in both contexts, in which behavioural factors such as distraction and lapses led to speeding, near-misses and eventually crashes. However, distinct contextual differences were also evident. School zones exhibited a broader and more context-sensitive behavioural structure, with speeding strongly influenced by situational factors, such as pickup and drop-off activity, as well as demographic and behavioural variables. In contrast, roadwork zones showed a more constrained, task-driven pattern, in which outcomes were primarily influenced by attentional factors and exposure. Near-miss involvement was consistently and strongly associated with driving experience across both contexts and acted as a key intermediate risk indicator. Subsequently, near-miss events were found to be strong predictors of crash risk in both environments. Additionally, driving experience, risk perception, and driving exposure influenced crash involvement in both environments. Overall, the findings highlight the importance of modelling mediated behavioural pathways and demonstrate that driver behaviour is highly context dependent. |
Paper 111
Impact of surrounding vehicles on cognitive load during work zone lane-changing manoeuvres
Room Location: B11.B2.101
View Authors
- Tharindu Ranawaka (Deakin University)
- Sajani Siriwardene (Deakin University)
- Davies Huw (Coventry University)
- Christophe Bastien (Coventry University)
- Ashim Debnath (Deakin University)
View Abstract
| Work zones pose significant challenges to drivers due to temporary lane closures, altered traffic patterns, and increased interaction with surrounding vehicles, often resulting in elevated cognitive load during lane‑changing manoeuvres. Understanding how cognitive load is influenced by driver characteristics and surrounding‑vehicle configurations is critical for improving safety, however, limited knowledge exists in this area. This study investigates drivers’ cognitive load during work‑zone lane merging using eye‑tracking data collected from a driving‑simulator experiment involving 60 adult drivers, each exposed to 16 lane‑closure scenarios with varying traffic conditions. A multilevel mixed‑effects linear regression model was used to identify factors influencing cognitive load, measured using pupil diameter. Results showed that cognitive load is significantly influenced by the presence and types of different surrounding vehicles, driver demographic characteristics, driving experience, and merge location. Eye‑movement measures were also strongly associated with cognitive load, with higher fixation ratios and increased saccade frequency corresponding to lower pupil‑based load, and higher saccadic peak velocities associated with increased load. Findings provide new insights into how surrounding‑vehicle interactions influence drivers’ cognitive load during work‑zone lane‑changing manoeuvres. |
Paper 135
A random parameters bivariate extreme value model for 1 estimating rear-end crash risk by injury severity
Room Location: B11.B2.101
View Authors
- Jinbao Zhang (Queensland University of Technology)
- Ninad Gore (Queensland University of Technology)
- Shimul Md Mazharul Haque (Queensland University of Technology)
View Abstract
| Conflict-based traffic safety assessment methods have increasingly accounted for unobserved heterogeneity in crash risk estimation; however, incorporating evasive driver behaviour into multivariate extreme value modelling frameworks remains relatively underexplored. This study develops a bivariate random parameters extreme value model to estimate rear-end crash risk by injury severity at signalised intersections using evasive-action-based traffic conflict measures. Video data were collected from five signalised intersections in Queensland, Australia. The proposed framework combines Modified Time-to-Collision (MTTC) and Deceleration Rate to Avoid a Crash (DRAC) with the expected post-collision velocity difference (Delta-V) to jointly model crash frequency and injury severity outcomes. The random parameters bivariate framework was evaluated against equivalent fixed parameters specifications. The findings suggest that accounting for unobserved heterogeneity improves overall model performance and enhances crash frequency estimation. The results indicate that the rear-end crash risk is positively associated with the frequency of rear-end conflicts per signal cycle, while the proportion of right-turning vehicles influences both crash risk and injury severity. Among the traffic conflict measures examined, the DRAC-based model demonstrated comparatively stronger predictive capability and generated crash estimates that aligned more closely with observed crash frequencies by severity level than the MTTC-based model. Overall, the study highlights the potential of integrating evasive-action-based traffic conflicts within a bivariate extreme value modelling framework to improve the behavioural representation and predictive performance of conflict-based safety assessment at signalised intersections |
Paper 34
Exploring links between community transport and social capital, two case studies of regional Australian community transport providers
Room Location: B11.B2.103
View Authors
- Sandra Wilson (Monash University)
- Graham Currie (Monash University)
- Alexa Delbosc (Monash University)
View Abstract
| There is a consistent purpose of community transport to meet the local needs for those that don’t have access to or can’t use public transport or a private car. Delivering a community transport service relies on volunteers, trust, attitudes and social networks. There is a growing understanding that social capital may influence or be influenced by community transport, given that key elements of social capital are networks, trust and mutual benefit. However, there has been limited empirical research measuring links between social capital and community transport service. This paper summarises the findings from interviews with two community transport providers in regional Australia. Social capital of the main participants involved in each community transport service are measured. Influential mechanisms, context or attributes between social capital and community transport are outlined. |
Paper 86
Inclusive shared micromobility in the digital age: A cross-country study
Room Location: B11.B2.103
View Authors
- Yuting Zhang (The University of Sydney)
- Helena Titheridge (University College London)
- Aitan Medeiros Militão (Queensland Institute of Technology)
- Daniel Oviedo Hernandez (University College London)
- John D Nelson (The University of Sydney)
- Corinne Mulley (The University of Sydney)
View Abstract
| Micromobility devices such as bikes, e-bikes, and e-scooters have emerged as a growing transport mode. Shared micromobility, defined as rental systems that provide access to these devices, offers opportunities for pursuing sustainable transport objectives, particularly by supporting first- and last-mile connections to public transport. This research examines the impact of technology-enabled shared micromobility programmes on social inclusion, drawing on data from an online panel survey (N = 1,033) and expert consultations (N = 17) conducted in selected jurisdictions in Australia and England with differing levels of maturity and experience. The findings show that technology functions not only as a central component to the operation of shared micromobility programmes but also as an enabler and, simultaneously, as a significant but often underestimated barrier to inclusive mobility. While a range of best practices aimed at mitigating the adverse inclusionary impacts of technology-enabled shared micromobility internationally were identified, few have achieved meaningful uptake. Cross-jurisdictional comparison in this study reveals that differences in dominant vehicle types, governance approaches to programmes, and urban socio-spatial distributions play important roles in shaping micromobility use scenarios. These findings contribute to the development of policy and planning strategies to support more inclusive urban mobility systems. |
Paper 94
The bicycle as a social intervention: Visualising patterns of use to understand the impact of free share eBikes.
Room Location: B11.B2.103
View Authors
- Robbie Napper (Monash University)
- Chuan Khoo (Monash University)
- Chris Arnott (RideKola)
View Abstract
| Data visualisation is used to study the operational and social characteristics of trips made using a free shared eBike based in one of three social and affordable housing developments in Melbourne, Australia. Using time and geographical data, we extrapolate likely cycling routes and overlay this information with known geographic factors such as planning zones and speed limits. Free, hub-based ebikes located in residences are not common in Australia. Our analysis finds a wide variety of uses for the bicycles, not aligned with typical Australian bicycle uses of recreation and commuting. Short, multi-stop journeys are made on the bikes, in similar proportion to commuting trips, however the latter are not at typical peak periods. The research shows highly flexible application of the bikes across time and geography, suggesting that they are adopted into uses aligning with the purpose of social and affordable housing, beyond the two dominant uses of recreation and commuting typically seen in Australia. |
Paper 16
Socio-demographic influences on the transit transfer penalty: A systematic review
Room Location: B11.B2.103
View Authors
- Shaoyan Zou (Monash University)
- Graham Currie (Monash University)
- Rumana Sarker (Monash University)
View Abstract
| The act of transferring between public transport services remains one of the most significant deterrents to ridership. This paper presents the first systematic review of empirical evidence of how transfer penalties vary by socio-demographic groups. It also explores causal factors which influence these variations. After data screening, 59 separate studies were used to build the review. Results find transfer penalties are generally larger for women vs men (female-to-male ratios ranged 1.02×–5.38×; mean 2.17×), for non-commuters vs commuters (non-commuter-to-commuter ratios 1.09×–2.26×; mean 1.63×) and for low vs high income groups (low income mean ratio was 3.70×). Results for age (young vs old) and for those with disability were mixed and are hence inconclusive. Causal factors for socio-economic differences in transfer penalties identified 10 individual causes grouped into psychological and cognitive factors, physical and environmental factors and operational and spatiotemporal factors. An evidence weighting methodology is developed to synthesise the literature into a consistency matrix with findings identifying ten causal factor groups and thirteen demographic segments where transfer penalties vary. Implications for future research and practice are discussed. |
Paper 6
Exploring the scale and variation of private vehicle and non-private vehicle trips generated by residential developments in Australia and New Zealand
Room Location: B11.B2.104
View Authors
- Chris De Gruyter (RMIT University)
- Supun Perera (The University of Sydney)
- Steve Pemberton (RMIT University)
View Abstract
| Trip generation studies for land use developments have traditionally focused on private vehicles, with little attention given to non-private vehicle trips, such as public transport, walking and cycling. In addition, there is little understanding of how site-specific and local area characteristics influence non-private vehicle trips. This research explores the scale and variation of private vehicle and non-private vehicle trip generation using a unique dataset of 169 residential developments in Australia and New Zealand. The results show significant variation in trip generation rates and mode shares, with non-private vehicle trips across all sites ranging from almost 0% to 80% of all trips generated, at an average of around 40%. Through regression modelling, lower private vehicle trips are found to be significantly associated with higher density housing and lower off-street car parking provision, while non-private vehicle trips are significantly associated with access to higher frequency bus and train services. The Decide and Provide approach, with a basis in scenario planning, is also used to help demonstrate how the findings can be applied in practice to better accommodate uncertainty in trip generation. The research findings highlight the need for a fundamental shift in current trip generation practice to ensure that all modes of transport are adequately considered in planning for new residential developments. |
Paper 33
Understanding kerbside EV charging behaviour: Evidence from utilisation and land use profiles in NSW and Victoria
Room Location: B11.B2.104
View Authors
- Christina Chiu (University of Technology Sydney)
- Martin Buil (Centrale Méditerranée)
- Kasun Wijayaratna (University of Technology Sydney)
- Kriti Nagrath (University of Technology Sydney)
- Scott Dwyer (University of Technology Sydney)
- Michelle Zeibots (University of Technology Sydney)
View Abstract
| Kerbside electric vehicle (EV) charging is emerging as a critical component of urban charging infrastructure, particularly for households without access to off-street parking. However, empirical evidence on how these chargers are used in Australian real-world settings remains limited. This study presents a data-driven analysis of kerbside EV charging behaviour using operational data from a network of chargers across New South Wales and Victoria, Australia. Charger utilisation, session characteristics, and reliability metrics are analysed and linked to surrounding land use profiles derived from ABS mesh block data. Performance metrics of chargers based on utilisation patterns (e.g. solar vs overnight use) and land use clustering (e.g. residential, mixed-use, high street) have been produced. Results indicate that kerbside charging is predominantly used for shorter-duration opportunistic charging rather than long-duration overnight charging. High-utilisation chargers are typically associated with residential areas but also exhibit mixed temporal usage patterns, suggesting multiple user types and charging purposes. Reliability, measured through session-based and time-based failure rates, is shown to influence user behaviour and potential charger port switching. The findings provide new insights into the role of kerbside charging within the broader EV ecosystem and highlight implications for infrastructure planning, level-of-service frameworks, and policy design. This study contributes one of the first empirical assessments of kerbside charging behaviour in Australia and offers a transferable analytical framework for future evaluations. |
Paper 59
Rethinking parking standards for larger vehicles: Empirical evidence to inform the revision of AS/NZS 2890.1
Room Location: B11.B2.104
View Authors
- Mitchell Elbourne (University of Technology Sydney)
- Kasun Wijayaratna (University of Technology Sydney)
View Abstract
| The increasing size of passenger vehicles in Australia, particularly the growth of sport utility vehicles (SUVs) and dual-cab utilities, is placing pressure on existing parking infrastructure designed to earlier vehicle standards. Although there is an active initiative to update parking standards, the current design guidance has not been formally updated since 2004 (AS/NZS 2890.1:2004), raising questions about its ongoing suitability. This study evaluates the adequacy of current parking design standards through a three-phase empirical investigation. First, a historical analysis of vehicle dimensions (2018 – 2024) was used to reassess design vehicle parameters. Second, an observational survey across six off-street parking facilities in Sydney quantifies the prevalence of dimensional violations relative to current standards. Third, a series of digital redesign scenarios assesses the trade-offs between updated design dimensions, operational efficiency and parking capacity. Results indicate that current design vehicles no longer represent the contemporary fleet, with observed exceedances in vehicle width and turning requirements since approximately 2020. Field observations show violation rates of 6–28% for width and 4–12% for length, suggesting systematic misalignment between design assumptions and real-world usage. Redesign scenarios demonstrate that accommodating larger vehicles reduces parking capacity by approximately 3 to 15%, but can improve manoeuvrability and operational performance, particularly under angled parking configurations. The findings highlight a need to update parking design standards and provide empirical evidence to inform future revisions of AS/NZS 2890.1. |
Liton Kamruzzaman (Monash University)
Saeid Adli (Monash University)
Graham Currie (Monash University)
Le Hai Vu (Monash University)
Paper 72
A two-level land use and transport interaction framework for evaluating sustainable city size: A proposed approach for Australian high-speed rail assessment
Room Location: B11.B2.104
View Authors
- Dong Xiao (Monash University)
- Liton Kamruzzaman (Monash University)
- Saeid Adli (Monash University)
- Graham Currie (Monash University)
- Le Hai Vu (Monash University)
View Abstract
| City size can be shaped by the interaction between transport infrastructure, land use, and spatial economic structure. Conventional fine-grained urban Land use and Transport Interaction (LUTI) models, however, are not directly suited to assess corridor-scale infrastructure change or development policies, where key impacts occur across long-distance and intercity travel markets. A two-level LUTI framework is therefore developed, consisting of an intercity component for redistribution between major city-regions and an intracity component for redistribution within regions. Using Australian east-coast high-speed rail as the motivating case, the framework is designed to connect intercity accessibility change with intracity land use response. By separating intercity and intracity mechanisms while linking them through accessibility and redistribution effects, it addresses the spatial and behavioural mismatch between fine-grained urban LUTI modelling and corridor-scale infrastructure assessment and provides a clearer basis for future scenario testing. The resulting structure supports the assessment of high-speed rail as a city-shaping intervention and provides a foundation for future empirical analysis of sustainable city size in Australia. |
Paper 50
Behavioural insights from an international survey on mobility
Room Location: B11.B3.103
View Authors
- Adri van der Mescht (Ipsos)
View Abstract
| This paper examines key insights from the Ipsos Mobility Report, aligning with the theme 'Transforming Transport Futures: Integrating People, Place, and Technology.' The analysis highlights the deep-rooted reliance on cars among Australians, where car ownership is among the highest globally. This dependency calls for an examination of socio-cultural factors influencing car use and areas where strategies can encourage a shift towards sustainable transport solutions. The report reveals some generational behaviours that may be encouraging for the take-up rate of public transport usage among younger demographics, like Millennials and Gen Z. These groups demonstrate a higher proclivity for public transport, creating opportunities for interventions that address convenience and accessibility concerns. By understanding these dynamics better, policymakers can incorporate them into strategies to enhance public transport uptake and promote sustainable transportation. Australians express awareness of environmental issues in their transport preferences, though the reality is that this isn't always necessarily reflected in behaviour. Support for sustainable transport options like public transport is tempered by socio-economic factors and convenience, leading to continued reliance on personal vehicles. Road safety perceptions indicate verbal support for stricter traffic laws and dedicated cycling lanes. These are underpinned by demographic variations, guiding potential educational initiatives on road safety and speed management. Additionally, as vehicles become smarter and more connected, addressing data privacy concerns is crucial. Trust levels in automakers and governments regarding data protection are moderate, pointing to the need for robust regulations and consumer awareness. Through these insights, the paper offers data-driven insights for policymakers that can help in the formulation of strategies that prioritise accessibility, safety, and sustainability. |
Paper 53
Analyzing electric vehicle owners’ future continuation intentions: A discrete choice modelling approach
Room Location: B11.B3.103
View Authors
- Bisma Khalid (University of New South Wales)
- Milad Ghasri (University of New South Wales)
- Alireza Abbasi (University of New South Wales)
View Abstract
| The transition to low emission transport is accelerating globally; however, limited evidence exists on how current electric vehicle (EV) owners evaluate future vehicle choices, particularly in the Australian context. This study examines 467 EV owners’ future continuation intentions using a Discrete Choice Experiment (DCE) across four vehicle types battery electric vehicles (BEVs), plug-in hybrid electric vehicles (PHEVs), hybrid electric vehicles (HEVs), and internal combustion engine (ICE) vehicles, grounded in Random Utility Theory. The Nested Logit results reveal distinct preference structures across EV and non-EV alternatives. Driving range and operating cost significantly influence choices in both nests, with stronger effects for non-EV alternatives, indicating greater sensitivity when EV owners consider switching away from EVs. Charging infrastructure is significant only within the EV nest, reflecting experience-driven preferences. Purchase price has a negative and significant effect in both nests, with a larger magnitude for non-EV alternatives, suggesting higher price sensitivity when evaluating non-EV options. Body type plays a strong role in EV choices, with significant positive preferences for hatchbacks and SUVs. For non-EV alternatives, hatchbacks and large SUVs are also positively valued, although the effects are smaller and less consistent. The inclusive value parameters support a valid nesting structure. Overall, EV owners priorities efficiency, vehicle characteristics, and infrastructure within EV choices, while exhibiting greater sensitivity to cost and range when considering non-EV alternatives. |
Paper 57
A scenario-based system dynamics framework for multimodal travel behaviour during mega-event
Room Location: B11.B3.103
View Authors
- Ishani Shehara Pitigala Liyana Arachchi (Queensland University of Technology)
- Shellee Franssen (Department of Transport and Main Roads)
- Geoffrey Walker (Queensland University of Technology)
- Shamsunnahar Yasmin (Queensland University of Technology)
View Abstract
| Mega events such as the Olympic and Paralympic Games represent the most demanding test of multimodal transport systems, requiring rapid and substantial mode share shifts within highly constrained timeframes. Despite this, existing transport planning approaches remain largely linear and mode-specific, failing to capture the feedback-driven, nonlinear dynamics that shape multimodal travel behaviour. As a result, policy interventions often address isolated components of the system, limiting their effectiveness in achieving sustained modal shift and long-term transport legacy outcomes. To address this gap, this study develops a scenario-led systems-thinking framework to analyse multimodal travel opportunities in mega-event contexts. Using a multi-phase mixed-methods design, the study integrates literature-based scenario development, expert focus group elicitation, and a structured follow-up survey assessing policy feasibility, multimodal influence, and mega-event priority across fifteen transport system variables. These inputs are synthesised into a Causal Loop Diagram model comprising thirteen feedback loops across three scenarios: Smart city mobility, Seamless first/last Mile, and Inclusive mobility for all. The results indicate that public transport reliability and interchange integration function as dominant system drivers, while digital tools and active travel infrastructure act as scenario-specific amplifiers. Mapping the Avoid–Shift–Improve framework onto feedback structures identifies congestion pricing, Olympic hub formalisation, and public transport frequency uplift as high-leverage interventions. Loop dominance analysis further reveals that event-driven demand creates a temporary window for reinforcing modal shift dynamics. Without a systems-based approach, mega-event transport planning risks relying on fragmented interventions that fail to produce lasting change. This study contributes a novel, scenario-sensitive CLD framework that enables planners to design integrated policy strategies addressing both event-time performance and post-event legacy. |
Paper 96
Multi-day travel, activity, and expenditure data collection for week-long ABM using passive tracking
Room Location: B11.B3.103
View Authors
- Maliheh Tabasi (University of New South Wales)
- Taha H. Rashidi (University of New South Wales)
View Abstract
| Activity-based models (ABMs) estimate travel demand by simulating individuals’ activities, their schedules, locations, and social contexts, thereby generating the trips associated with participation in out-of-home activities. As such, ABMs provide a valuable tool for assessing transport policies. However, most existing ABMs focus on a single typical day, which may not adequately capture the variability in individuals’ activity and travel patterns. Prior literature suggests that one week of activity and travel data is sufficient to represent interactions between in-home and out-of-home activities and to evaluate the impacts of relevant policies, such as working from home. In this paper, we present our effort to collect such multi-day data using passive tracking methods. We describe the data collection framework, and provide descriptive analyses of the resulting dataset. The dataset offers significant potential for addressing a wide range of research questions across multiple disciplines, particularly those related to travel behaviour and policy evaluation. |
Paper 13
Parcel locker usage frequency in Queensland, Australia
Room Location: B11.B3.101
View Authors
- Mostafa Khatami (University of Wollongong)
- Prithvi Bhat Beeramoole (Queensland University of Technology)
- Mohammad Mahdi Ahmadian (Queensland University of Technology)
- Alexander Paz (Queensland University of Technology)
View Abstract
| The rapid growth of e-commerce has led to increased demand for efficient last-mile delivery solutions, with parcel lockers emerging as a popular alternative to traditional home delivery. While previous research has focused predominantly on the optimal placement of parcel lockers, understanding the factors that influence consumer usage frequency remains underexplored. This study addresses this gap by examining what drives the use of parcel lockers in Queensland, Australia. Using survey data collected from residents across Queensland, we adopt a two-stage modelling framework that captures both the decision to use parcel lockers and the frequency of use among adopters. The analysis incorporates a range of explanatory factors, including socio-demographic characteristics (such as age, income, and education), spatial and accessibility-related attributes (including proximity to parcel lockers and urban form), behavioural factors (such as online shopping frequency and preferences), and perceptual considerations (including convenience, security, and reliability). By jointly accounting for these dimensions, the study provides a framework for understanding parcel locker adoption and usage behaviour. The findings offer insights for service providers, and policy-makers seeking to design and manage more effective parcel locker networks. Overall, parcel locker usage appears to depend less on regional context and more on alignment with users’ daily routines and preferences, underscoring the importance of user-centred design and motivating future research on behavioural dynamics and heterogeneity. |
Paper 90
Heavy vehicle crash causal factors, severity and location analysis in South Australia
Room Location: B11.B3.101
View Authors
- Sajani Siriwardene (Deakin University)
- Oskar Vickor (Deakin University)
- Ashim Debnath (Deakin University)
View Abstract
| Heavy vehicle (HV) crashes are typically associated with greater severity levels than light vehicle (LV) crashes due to the mass and operational characteristics of these vehicles. Consequently, the factors influencing HV crash occurrence and severity may differ from those of LVs. Despite ongoing HV safety initiatives, HV crash numbers in South Australia remain persistently high, warranting further investigation. This study examines the factors contributing to HV crash occurrence and severity using historical crash data from South Australia. To derive further insights into these factors, an ordered probit model was employed, while a spatial analysis was used to identify high-risk locations. The spatial analysis revealed nine high-risk locations in the Adelaide region, two of which are currently treated for safety upgrades. Based on these findings, targeted recommendations are proposed to improve safety outcomes at these locations. Furthermore, the model results indicated that HV crashes are more likely to occur on straight, level road segments under dry conditions and in well-lit environments, indicating that driver-related factors, particularly inattentiveness, play a significant role in HV crash frequency under otherwise favourable driving conditions. Even though the crash numbers are higher, these favourable conditions resulted in less severe crashes. Conditions such as horizontal curves, nighttime, and higher posted speed limits led to more severe crashes. In addition, driver impairment due to alcohol and drug use was identified as a critical contributor to increased crash severity. Overall, this study provides insights into the unique risk profile of HV crashes and highlights the need for targeted interventions addressing driver behaviour and location-specific risk factors. |
Paper 141
Impacts of electricity pricing on charging and inventory strategies at battery swapping stations in shipping
Room Location: B11.B3.101
View Authors
- Yubing Wang (Shanghai Jiao Tong University)
- Lei Dai (Shanghai Jiao Tong University)
- Siqing Guo (Shanghai Jiao Tong University)
- Mingyuan Yue (Shanghai Jiao Tong University)
View Abstract
| This study investigates the impact of electricity pricing on charging scheduling and battery inventory strategies at battery swapping stations for inland shipping. A mixed-integer linear programming model is formulated for a spatially decoupled charging–swapping network, where charging stations and swapping stations are separately located and connected through battery transportation. Electricity prices are treated as exogenous inputs that vary across time and location, enabling the analysis of fixed pricing, time-of-use (TOU) pricing, and an illustrative random-price scenario. Results from the Shanghai–Nanjing route show that battery procurement cost dominates the total system cost and is primarily driven by exogenous swapping demand and vessel schedules, whereas electricity cost is highly responsive to pricing signals. Under a wide TOU peak-valley spread, such as ±75%, electricity cost can be reduced by up to 72.1% through strategic shifting of charging activities to valley periods. In contrast, a marginal spread, such as ±10%, may increase total cost because limited valley discounts are insufficient to offset charging that must still occur during higher-price periods. The results highlight the role of battery inventory as an operational buffer that enables charging deferral and supports flexible responses to electricity price variations. These findings provide operational guidance for swapping station operators and inform tariff design for inland shipping electrification. |
Paper 150
Near-miss events as leading indicators of crash risk: A spatial-statistical analysis of greater Sydney using connected vehicle data (2023–2024)
Room Location: B11.B3.101
View Authors
- Javiera de la Carrera (Compass IoT)
View Abstract
| Road safety management traditionally relies on historical crash records, an inherently reactive approach requiring harm to occur before intervention. This paper presents a spatial-statistical framework leveraging connected vehicle telematics from Compass IoT to proactively identify crash risk in Greater Sydney, extending Grigorev et al. (2025). Using High-G near-miss events from 117,205 records across 2023–2024, and 37,723 TfNSW (Transport for New South Wales) crash records, Bivariate Local Moran’s I (LISA) is applied to a 400m grid across 12 urban SA4 zones (3,468 km², 22,885 cells). Global Moran’s I confirms significant spatial clustering of crashes (I=0.338) and near-misses (I=0.207, both p<0.001). LISA identifies 995 Low-Crash High-Near-Miss (LH) cells in 2023 — all with zero recorded crashes — of which 26.2% became confirmed crash blackspots (HH) by 2024, validating the prospective predictive power of telematics data. Negative Binomial regression (Pseudo R²=0.249) identifies braking-type near-misses (IRR=14.7) as the strongest crash predictor, substantially outperforming steering events (IRR=8.9). A one standard deviation increase in mean G-force severity (0.083g above the active-cell mean of 0.611g) is associated with 5.3% more crashes (IRR=1.053, p=0.030), confirming that near-miss severity complements frequency as a risk signal. SA4-level Spearman correlations confirm the near-miss to crash relationship across all zones (ρ=0.415–0.643, p<0.001), with the CBD exhibiting the strongest co-location. These findings demonstrate that connected vehicle telematics can identify high-risk locations before crashes occur, enabling a shift from reactive to proactive road safety management. |
5:20 – 6:30
Travel to Conference Gala Dinner
6:30 – 10:00
Conference Gala Dinner
Dockside