From truck driver awareness to obstacle detection: A tiger never changes its stripes

To inform the design of artificial intelligent systems on a mine site, incidents involving driverless haul trucks were evaluated to understand the risk implications of automation in its application. Safety-related incidents (n = 998) on a mine site in Western Australia (WA) were recorded to analyse events involving driverless and manually operated haul trucks since the operation began back in 2013. Truck incidents were evaluated and compared on their characteristics and investigation findings. From FY14 through to FY18, the incident frequency of manually driven haul trucks averaged 968 incidents per 1,000,000 hours driven, while the driverless trucks averaged 866. Driver awareness was the most frequent hazard associated with manually operated haul trucks, while haul road conditions (objects identified or not) were the most common hazard associated with automated haul trucks.

Data analysis demonstrates how driverless trucks transformed a mine site’s risk profile, rather than underpin the popular notion that automation eliminates the risks associated with surface mobile equipment. Therefore, risk management should focus on enhancing users’ knowledge of computer programming and machine learning techniques that is driving the most progress in industry to-date. Such a focus would legitimise the current progress of artificial intelligence and highlight the residual workload of humans whose roles are transforming and adapting to the introduction of driverless technology.