How Do eHMIs Affect Pedestrians’ Crossing Behavior? A Study Using a Head-Mounted Display Combined with a Motion Suit

Round 1

Reviewer 1 Report

Although the starting point of this research comes from the summary of existing research, it is meaningful to investigate the effects of eHMIs on participants’ crossing behavior, and the method using a virtual urban environment by means of a head-mounted display coupled to a motion-tracking suit is also feasible and innovative. The operation of the experiment is thoughtful and the data processing is more rigorous. It is suggested to supplement brief explanation for following recommendations.

This paper refers that FBL is ambiguous. Whether the participants were explained what the meaning of the FBL color is before the experiment when studying the effects of FBL on participants’ crossing behavior? However, in the actual promotion, the meaning of FBL may be promoted like most current traffic rules, and is widely known. Under such different conditions, will the behavior of pedestrians change? The MVN data is low-pass filtered using a zero-phase 10th order Butterworth filter. Why is this filter suitable for this study?

Author Response

Although the starting point of this research comes from the summary of existing research, it is meaningful to investigate the effects of eHMIs on participants’ crossing behavior, and the method using a virtual urban environment by means of a head-mounted display coupled to a motion-tracking suit is also feasible and innovative. The operation of the experiment is thoughtful and the data processing is more rigorous.It is suggested to supplement brief explanation for following recommendations.

R: Thank you for your review. 

This paper refers that FBL is ambiguous. Whether the participants were explained what the meaning of the FBL color is before the experiment when sudying the effects of FBL on participants’ crossing behavior?

R: Participants were not informed about the meaning of the eHMIs. We have amended the Methods section.

However, in the actual promotion, the meaning of FBL may be promoted like most current traffic rules, and is widely known. Under such different conditions, will the behaviour of pedestrians change?

R: This is true; the front brake lights may not be ambiguous anymore after training/learning. This is now emphasized in the Discussion, together with literature support.

The MVN data is low-pass filtered using a zero-phase 10th order Butterworth filter. Why is this filter suitable for this study?

R: The data were filtered according to Schreven et al. (2015), who showed that the optimal cut-off frequency for human motion is about 8 Hz. We enhanced the Data reduction section with a description of our filtering steps.

Reviewer 2 Report

The abstract requires major revision. Especially, the first few sentences are not connected and confuse the readers about the study's purpose and methods. The authors should present the details of the participants without exposing their PIIs.  A possible way could be to present these in a table. "train of cars"--this phrase is used numerous times in the paper. Is it the platoon of cars? If so, please use this term in the paper.  The platoon of cars is the most used term in the case of AV/CV. The experimental design and implementation are very straightforward. The authors need to provide more details regarding the novelty of their approach rather than just adopting the ones from other articles. The major concern regarding the findings reported is statistical testing. Given a small sample size, t-tests are performed, which is correct. However, it implies that the data follow a Normal distribution, which we don't know. It is recommended to perform some non-parametric tests to verify whether the results from t-tests can be accepted. Figs. 3,4,5: Why is the significance level presented in the log-scale? Fig. 7 has a very bad resolution. The major weakness of the paper is in the Discussion section.  The authors need to analyze their findings more critically with that of the others. Furthermore, some recommendations are needed.  Without that, it's just another technical exercise. There are several proof-reading errors. For instance, line-291, line-335, line-467, etc.

Author Response

(x) Moderate English changes required

R: We have done a careful proofread of the manuscript, resulting in a number of spelling corrections.

The abstract requires major revision. Especially, the first few sentences are not connected and confuse the readers about the study's purpose and methods.

R: We have rewritten the first few sentences of the abstract and now provide more context and point to a research gap.

The authors should present the details of the participants without exposing their PIIs.  A possible way could be to present these in a table.

R: We have added information about walking frequency, age range, nationalities, and recruitment.

"train of cars"--this phrase is used numerous times in the paper. Is it the platoon of cars? If so, please use this term in the paper.  The platoon of cars is the most used term in the case of AV/CV.

R: Amended.

The experimental design and implementation are very straightforward. The authors need to provide more details regarding the novelty of their approach rather than just adopting the ones from other articles.

R: In the Discussion, we now reflect on the innovation that participants could see their avatar.

The major concern regarding the findings reported is statistical testing. Given a small sample size, t-tests are performed, which is correct. However, it implies that the data follow a Normal distribution, which we don't know. It is recommended to perform some non-parametric tests to verify whether the results from t-tests can be accepted.  

R: We have repeated all statistical tests with signed-rank tests. The results are provided in the supplements. The results for the signed-rank tests do not alter our conclusions. The text in the Methods has been clarified.

Figs. 3,4,5: Why is the significance level presented in the log-scale?

R: In Manhattan plots, the logarithm of the p-value is usually shown. This is done for discernibility of p-values. For example, a p-value of .01 and a p-value of .001 are hard to distinguish on a linear scale, despite the fact that the effect size is notably different. We have added a clarifying sentence to the Methods.

Fig. 7 has a very bad resolution.

R: Replaced with a high-resolution version.

The major weakness of the paper is in the Discussion section.  The authors need to analyze their findings more critically with that of the others. Furthermore, some recommendations are needed.  Without that, it's just another technical exercise.

R: We have substantially extended the discussion and now cite more research. More specifically, we reflect on the fidelity of the setup and make recommendations about whether or not it is necessary to use a high-fidelity measurement setup for evaluating eHMIs, see also R10.

There are several proof-reading errors. For instance, line-291, line-335, line-467, etc.

R: These formatting errors have been corrected.

Round 2

Reviewer 2 Report

Please do another round of proof-reading.