Helping improve level crossing safety through data-led insights

Rail level crossing crashes are infrequent, but when they occur the consequences can be severe for road users, rail operations and surrounding communities. Traditional safety assessments rely heavily on site inspections, which can be resource-intensive and difficult to undertake regularly across a large, geographically dispersed network.

TfNSW wanted to strengthen its ability to identify higher-risk locations across the network, particularly sites where no incidents had occurred but operating conditions indicated greater exposure to risk. To do this, it required a more scalable approach that could complement existing assessment methods and provide broader network-wide visibility of risk.

Working alongside TfNSW and supported by an Australian government Research and Innovation grant under the Regional Australia Level Crossing Safety Program, we helped develop an approach that enhances existing level crossing safety assessments. We incorporated driver behavioural insights based on data from connected vehicles - those that are equipped with Internet of Things (IoT) technology - and risk forecasting techniques.

TfNSW already holds extensive road and rail data about its level crossing network. We helped bring that information together with new insights into driver interactions with level crossings. Linking operating conditions, driver behaviours and historical safety outcomes to predictive insights has provided a richer, scalable evidence base for risk assessment and decision-making.

This work is enabling TfNSW to assess level crossing locations more regularly and consistently across the network and strengthen its understanding of where further investigation or safety interventions could deliver the greatest benefit.

Key elements of our approach included:

• Analysing connected vehicle data to provide new behavioural safety insights that are not typically captured through traditional site-based assessments
• Bringing together existing road and rail data with driver behaviour insights relating to more than 1,300 public level crossings across NSW
• Using risk forecasting techniques to identify locations with higher relative safety risk and improve predictions of where serious incidents are more likely to occur
• Supporting existing assessment processes with additional insights and a more scalable evidence base that enables cost-effective statewide analysis
• Enabling network-wide assessment without requiring a physical visit to every crossing

The project demonstrates how operational data can deliver greater value when combined with behavioural insights and advanced analytics. The approach provides a broader understanding of risk than traditional methods alone. This enables earlier identification of locations that may warrant further review and more targeted consideration of treatments such as signage, signals and barriers.

The ability to assess more than 1,300 public level crossings using a consistent methodology also supports improved visibility of risk across regional areas, where site inspections may be conducted less frequently. This provides a foundation for more proactive risk management, helping agencies prioritise resources and focus safety investment where it can have the greatest impact.

Simon Hunter, Chief Transport Planner - Transport for NSW, said, "Every crash at a level crossing is one too many. That’s why we’re harnessing technology and data with the aim of improving risk assessments. The learnings we have obtained through this project provide another potential tool to improve level crossing safety, alongside infrastructure investments, education about safe behaviour and innovative research and trials."