Picture of a steering wheel and car interface

New insight into automated driving

​Previous research on automated driving is mainly based on simulator studies. Alberto Morando’s research shows the real-world effects of vehicle automation.
You often hear that automated vehicles have enormous potential to make driving safer and more comfortable. They can compensate for human limitations that may cause crashes. Alberto Morando has in his PhD project addressed some challenges surrounding the safety implications of automated driving.

“An important question is then whether assistive automation may give drivers the false impression that their attention to the road is no longer important. In fact, if automation fails when the driver is not attentive, a crash may happen.” says Alberto Morando.

The aim of his research was to make automated vehicles safer. He did so by studying the theory of attention, which explains how humans perceive and interact with the environment and by measuring how drivers naturally behave when using automated functions during their routine driving.

“We found that drivers looked less to the road ahead when the vehicle had assistive automation compared to when the vehicle was manually driven. This result seems to suggest that automation may, in fact, compromise safety”.

However, drivers were successful at changing their behavior according to the uncertainty of the driving context, e.g., presence of other vehicles, and light conditions, independently of automation.

“Our findings indicate that today’s automated systems may not reduce drivers’ ability to react to hazards on the road. Drivers are attentive at critical points and can respond to critical situations”.

The novelty of his research is in the use of large, real-world driving data and new methods for data analysis. Most of the previous research was based on simulator studies; the analysis showed the real-world effects of vehicle automation, which were unknown. He and his co-authors also applied sophisticated methods for data analysis, such as Bayesian methods. These methods allowed to describe in greater detail than before the effects of automation on driver behavior.

The outcome of his research have implications for assessment and testing of automated vehicles. The findings are provided in a format that can be used as a reference for driver modelling and driver models, implemented in computer simulations, that are becoming integral part of the design of safer vehicles.

“For example, driver models can predict what are the safety consequences of introducing a new vehicle system on the road. Our results can also be used to inform the design of real-time systems. Such system can provide continuous feedback to the driver and support safe driving.” 
Read more  

Published: Tue 19 Mar 2019. Modified: Wed 20 Mar 2019