Human-Centered Safety Metrics for Autonomous Trucks
Heavy-duty autonomous vehicles (HD-AVs), especially in freight, promise efficiency and safety gains — but they also bring new risks, particularly around human-system interaction. Our forthcoming report introduces a risk-informed framework for evaluating safety in HD-AV operations. The methodology builds on structured risk assessment approaches initially developed for driverless passenger cars, and adapts them to trucking by emphasizing the evolving roles of safety drivers and remote operators.
The research develops a set of human-centered safety metrics that capture the interactions between drivers and operators and ADS-equipped trucks. These metrics track issues such as:
Control transitions between automation and human drivers
System alerts and operator responses
Fallback events when automation fails
Trust and workload in human-ADS interactions
The approach combines quantitative indicators (e.g., takeover success rates, frequency of unacknowledged alerts) with qualitative insights from interviews and task analyses. Importantly, these metrics can be monitored continuously — through simulations, testing, and real-world operations — offering a proactive way to detect risks before incidents occur.
By centering safety evaluation on human-ADS interactions, this research provides regulators, developers, and operators with tools to design safer systems, guide training programs, and inform future safety standards. As autonomous trucking scales, these methods will be critical for ensuring that automation supports rather than replaces human oversight.