Effects of E-mirror position and size and lane changing direction on driver eye-off-road time, mental workload, and preference

Abstract

E-mirrors, mirror-less systems using in-vehicle visual displays and images taken by cameras, should be configured in a way that can minimize driver confusion and improve driver safety and preference. The objective of this study was to examine the effects of E-mirror position and size and lane changing direction on driver eye-off-road time, mental workload, and preference. Twenty (19 males and 1 female) young individuals with a mean (standard deviation; SD) age of 24.7 (2.2) years participated in this study. All participants were recruited from a university population and reported no musculoskeletal diseases. They had a valid driver license and over 2-year driving experience with a mean (SD) of 4.6 (2.0) years. This study used a 3-way (4 (E-mirror position; P) × 3 (Emirror size: S) × 2 (lane changing direction; D)) full factorial design. P considered the result of a focus group interview (FGI) carried out by the authors (P1: next to the sides of the room mirror), two recommended E-mirror locations (P2: the right side of the left A pillar bottom and the center of the dashboard top, P3: near the A pillar bottoms), and conventional outside mirror positions (P4). Two 8-inch tablet PCs with a 16:9 aspect ratio were used as E-mirrors. S included H6 (6 cm high; reflecting the mean size of inside rear-view mirrors (with a range of 6-13 cm)), H8 (8 cm; reflecting the mean preferred mirror size (8.5cm) from the FGI), and H9.7 (9.7 cm; reflecting the mean size of outside mirrors). D included changing to the right (DR) and left (DL) lanes. The eye-off-road (EOR) time was the time spent not looking at the road ahead during the lane changing task. The NASA-TLX questionnaire with six items was used to assess drivers’ mental workload on a 0-100 scale for each lane changing trial. Driver preference was verbally rated on a 9- point Likert scale (1: do not prefer this configuration at all, 3: do not prefer this configuration, 5: neutral, 7: prefer this configuration, 9: prefer this configuration the most) for the question of “How much do you prefer to use this E-mirror configuration (position and size) compared to conventional outside mirrors?”. Eye movement was measured using a goggle-type eye-tracking device. A driving simulator was used to perform lane changing tasks. Before the experiment, each participant was informed of the E-mirror concept and practiced driving the driving simulator until they familiarized themselves with all 12 E-mirror configurations (treatments). For each configuration, each participant made two lane changes (DR and DL), with their eye movement measured. Mental workload was rated for each trial, whereas preference was rated for each configuration. A 3-way analysis of variance (ANOVA) was used to analyze the main and interaction effect of EOR time, mental workload, and preference. If a main or interaction effect was significant, Tukey’s honestly significant difference test was used for post-hoc pairwise comparison. Significance was concluded when p < 0.05. The interaction effect of P×D for mental workload was significant (p = 0.017). The mean (SD) mental workload was lowest for P3× DL (34.7 (18.9)) and highest for P1× DR (46.9 (18.2)). The effect of P and D were significant for EOR (p < 0.0001 and p < 0.0001). The P levels were split in two groups (P2 and P4-P3-P1). The mean (SD) EOR time was shortest for P2 (4.2(1.5) s) and longest for P1 (5.1(1.6) s). The mean (SD) EOR time was shorter for DL (4.4 (1.6) s) than for DR (5.1 (1.7) s). The effect of P and D were significant for mental workload (p = 0.001 and p = 0.032). The P levels were split in two groups (P3-P2-P4 and P2-P4-P1). The mean (SD) mental workload was lowest for P3 (34.8 (18.0)) and highest for P1 (42.5 (22.0)). The mean workload (SD) was lower for DL (36.6 (20.4)) than for DR (39.6 (20.4)). The effect of P was significant for preference (p < 0.0001). The P levels were split in two groups (P2-P3-P4 and P1). The mean (SD) preference was highest for P2 (6.1 (1.83)) and lowest P1 for 4.6 (2.1). The effect of S was significant for preference (p < 0.004). The S levels were split into two groups (H9.7-H8 and H8-H6). The mean (SD) preference was highest for H9.7 (6.1(2.0)) and lowest for H6 (5.9 (2.1)). Thus, the effects of P for the safety-related factors (EOR time and workload) and the safety-unrelated factor (preference) were all significant, whereas the effect of S was significant for the safety-unrelated factor (preference) only and the effect of D was significant only for the safety-related factors (EOR time and workload). P2 provided the shortest EOR time and highest preference. Although P2, P3, and P4 were in the same group, P3 required 6.7% lower mental workload compared to P2. We suspected that spatial mapping would be more natural with P3 (vs. P2) because the right Emirror of P3 (vs. P2) is closer to the right lane. The size of Emirror significantly affected preference only. Similarly, though not significant (p = 0.65), The EOR time decreased as the display size increased (H6: 4.9s, H8: 4.7s, H9.7: 4.6s) in the current study. Although safety-related effects of S were not significant (p ≥ 0.35), larger E-mirror sizes appear desirable. DR showed a higher mental workload and a longer EOR time compared to DL. Several factors could have contributed to these results. For DR, the target lane and passenger-side Emirror are farther away from the driver, hence leading to decreases in the perceived display size and brightness. Overall, P2 is recommended for the EOR time, P3, P2, and P4 are recommended for the mental workload (the mean mental workload was lowest for P3), and P2, P3, and P4 and Emirror size ≥ H8 are recommended for preference (the mean preference was highest for P2 and for H9.7). Therefore, Emirrors ≥ H8 should be placed at P2 or P3.