The presence of a passenger can affect driver safety positively or negatively. Therefore gaining a better understanding of the nature of this interaction with a passenger is crucial. Additionally, drivers and passengers acquire nearly identical visual information from the driving scene. As a result, synchronised behaviour may occur. The driver-passenger synchrony may provide additional information on driver-passenger interactions. This led to the aim of the current study: The identification of synchrony in head orientation and movement between drivers and passengers for various road types and cornering events. This study was based on real-world driving data. The head angle (orientation) and angular velocity (movement) of all three principal axes (pitch, roll, and yaw) were analysed for synchrony.

The following methods were applied for the identification of driver-passenger synchrony. The head angles of drivers and passengers were detected using OpenFace 2.0, a video-based pose estimation method. Second, the windowed cross-lagged correlation (WCLC), a linear approach for synchrony identification, was applied. Next, two distinct facets of synchrony were measured. The first facet was determining the frequency of synchrony derived by the peak-picking algorithm developed by Altmann (2013). The frequency of synchrony was summarised as the percentage of synchrony that occurred over the measured road segment. The second facet of synchrony studied was the strength of synchrony with the help of the peak-picking algorithm developed by Boker et al. (2002). This peak-picking algorithm searches for the maximum peak correlation at every window of the WCLC. Then all maximum peak correlations were averaged into a single mean peak correlation per road segment or cornering event. Since behavioural synchrony could appear by chance, the results of the peak-picking algorithms were compared to pseudosynchrony. The first hypothesis states that it is possible to distinguish detected synchrony from pseudosynchrony of driver’s and passenger’s head angle and angular velocity. Secondly, it is hypothesised that more synchrony is detected in the urban road type than outside built-up areas- and highway road types.

The results showed that differentiation between the detected synchrony and pseudosynchrony could be made for the analysis on the total route for almost all head angles and angular velocity for the frequency- and strength of synchrony. Furthermore, the analysis on the different road types, revealed that the identified synchrony could be differentiated from pseudosynchrony for almost all urban road segments. The findings on the other road types were mixed. It can be concluded from these research results that synchronisation has been identified for the urban road type. The road types where the detected synchrony was not significantly different from pseudosynchrony could indicate that there was no synchrony present or that the applied methods were unable to capture the driver-passenger synchrony. Synchrony identification in cornering events revealed no observable patterns. As a result, no conclusions on the cornering events could be formed. The second hypothesis could only be tested for one condition, since the identified synchrony could not be differentiated from pseudosynchrony for all the required road types for the second hypothesis. The result showed no significant difference between the urban road type compared to the built-up areas and highway road types for the pitch–angular velocity.

According to this study, drivers and passengers exhibited synchronisation in their head orientations and movements along particular road segments. However, more research is needed to truly comprehend the synchronised behaviour of drivers and passengers. A good place to start is for a study that looks into the relation between synchronous behaviour and the impact on driver safety as a result of the passenger's presence.