Traditional road safety assessments require an unusually large number of crashes to accrue before transport infrastructure can be diagnosed for safety. In addition, police-reported crash data faces significant shortcomings, such as under-reporting, low sample means, limited behavioural information, and omitted variable bias. As such, traditional road safety models do not provide a complete picture of crash risk and are not capable of offering real-time road safety information. In this regard, Artificial Intelligence (AI)-based video analytics, together with traffic conflict techniques, offer a viable alternative for real-time assessment of transport infrastructure. However, without considering the theoretical foundation of traffic conflicts and crashes, the estimation of crash risks might be misleading. This talk will present recent developments in traffic conflict techniques to estimate crash risks from traffic conflicts. This talk will show how AI-based video analytics have been used to assess crash risk at transport facilities and demonstrate some use cases for real-time safety assessments and before-after evaluation of engineering treatments.
Is virtual reality the next frontier for traffic incident management capability and response coordination development?
June 28, 2024
Queensland state-controlled road users are no stranger to disruptions like vehicle crashes on the network. In managing disruptions, the Department of Transport and Main Roads
