Do We Trust Human Drivers More? Researchers Study Passenger Reactions in Self-Driving Cars
Researchers from the University of Szeged, in collaboration with experts from HUN-REN and the Budapest University of Technology and Economics, have investigated how passengers respond to unexpected situations in self-driving cars. Their findings have been published in Nature.
Previous physiological studies have shown that passengers experience lower levels of anxiety in cars driven by humans compared to autonomous vehicles. The aim of the current study was to investigate how passengers respond to unexpected road events in real-world scenarios.
The researchers conducted tests with 41 volunteers at the ZalaZONE Automotive Proving Ground. Each participant completed two approximately three-minute, two-kilometre journeys: one with a professional driver, and one in autonomous mode. Passenger responses were recorded using an electroencephalography (EEG) device and a specialised eye-tracking headset that captured eye movements, head movements, and blink frequency.
During the journey, participants encountered unexpected obstacles: dummies representing a deer and a child appeared on the road. The vehicle travelled at a maximum speed of 60 km/h. These events were designed to simulate scenarios that did not necessitate emergency manoeuvres but did require swift route correction.
The most notable finding from the EEG analysis was that participants preferred being in a car driven by a human over autonomous mode. In autonomous mode, they exhibited wider eye and head movements, along with a reduced blink frequency during unexpected events. However, the differences in eye movements were smaller than those observed in previous studies—likely because the steering, acceleration, and braking of the autonomous system had been modelled very closely on human driving behaviour.
The researchers concluded that measuring passengers’ physiological responses could support the wider acceptance of new modes of transport—especially in contexts where developers need to ensure that user expectations are being met.