A Robust Methodology for Dynamic Proximity Sensing of Vehicles Overtaking Micromobility Devices in a Noisy Environment

Round 1

Reviewer 1 Report

This study explores the implementation of a cost-effective, automated system  to accurately measure the passing distance and speed of vehicles near cyclists in urban settings. here some comments:

·        The abstract is too long, which avoids focusing on the contributions of the paper, authors are recommended to simplify it.

·        The authors are recommended to provide a more detailed comparison of their system's performance against existing methodologies for (LPD). This could include a discussion on the limitations of current methods.

·        In the introduction you should highlight the main contributions of the paper

·        the authors should implement control comparisons with existing measurement techniques could strengthen the methodology's validity.

·        To strengthen the conclusion, consider more explicitly linking the research outcomes to specific urban planning strategies

·        The authors should conduct a deeper analysis of the system’s potential limitations and error margins that would provide valuable context for interpreting its efficacy.

Minor editing of English language required

Author Response

We would like to thank the reviewers for their comments and suggestions. 

Comment 1. The abstract is too long, which avoids focusing on the contributions of the paper, authors are recommended to simplify it.

Reply 1: We appreciate the feedback. We have revised the abstract to make it more concise.

Comment 2. The authors are recommended to provide a more detailed comparison of their system's performance against existing methodologies for (LPD). This could include a discussion on the limitations of current methods.

Reply 2: Thank you for the suggestion. A detailed review of existing systems and methodologies have been added to the Introduction. 

Comment 3: In the introduction you should highlight the main contributions of the paper.

Reply 3: The main contributions of our paper have been highlighted in the introduction. 

Comment 4: The authors should implement control comparisons with existing measurement techniques could strengthen the methodology's validity.

Reply 4: Thank you for the suggestion. we have incorporated a new Table 2 that outlines the limitations of existing methods, accompanied by a detailed discussion in the Introduction section. Unfortunately, we encountered challenges in using the other systems described in the literature as controls for our experiments. These challenges arose primarily because many of these systems are not commercially available, and the descriptions provided in the literature did not offer enough detail to allow for accurate replication. This limitation notwithstanding, we are committed to further enhancing our methodology within the bounds of available resources and information.  

Comment 5: To strengthen the conclusion, consider more explicitly linking the research outcomes to specific urban planning strategies.

Reply 5: Thank you for your valuable suggestion. At present, our research has not directly tied to established urban planning strategies due to its pioneering nature in examining cyclist safety and micromobility systems. However, we share your vision and are hopeful that our findings will contribute to the development of innovative urban planning strategies. Specifically, we aim for our research to encourage enhanced data collection methodologies that prioritize cyclist safety. Moreover, we anticipate that our insights will pave the way for the implementation of cost-effective micromobility solutions.

Comment 6: The authors should conduct a deeper analysis of the system’s potential limitations and error margins that would provide valuable context for interpreting its efficacy.

Reply 6: We have expanded on Section 6, 'Limitations and Future Work,' to offer a more comprehensive examination of the system's potential limitations and error margins. This enhancement aims to provide readers with a clearer understanding of the context within which the system's efficacy should be evaluated, as well as to outline the direction of future research efforts required to address these limitations and improve the system's performance.

Reviewer 2 Report

I rate the peer-reviewed article highly. In it, the authors presented the results of their own research related to the movement of cyclists. In this research they installed specialized sensors in bicycles, which allowed them to assess the distance of other vehicles from these means of transport during an overtaking maneuver. They analyzed the data obtained, the results of which they presented very well in the reviewed article. The article therefore reads well.

The article is interesting and raises an important issue of road safety related to cyclists. I therefore think it should be published in the Applied Sciences journal.

However, I have a comment on the literature review. I believe that the authors could have included more up-to-date items. I therefore request that the article be improved in this regard.

Author Response

We would like to thank the reviewers for their comments and suggestions. 

Comment 1: However, I have a comment on the literature review. I believe that the authors could have included more up-to-date items. I therefore request that the article be improved in this regard.

Reply 1: In response to the reviewer's request, we have updated our literature review to include citations from an additional 17 recent articles that are of relevance to our study's focus. 

Reviewer 3 Report

1. 1. Recent research literature should be added, as this is a manuscript submitted in 2024, yet there are hardly any references from 2021 to 2024 in the paper.
2. 2. The contributions should be summarized again, given that this is a research paper.
3. 3. The font size of the axes in the figures should be increased to improve readability. Additionally, it is rare for studies to place the figure title on the top axis.
4. 4. Why is there a need to add a shaded background to equations (2), (4), and (5)?
5. 5. Why is there a need to consider vehicle speed? It seems that regulations only specify a minimum distance of 1.5 meters between bicycles and cars, without restricting their relative speed.
6. 6. If clustering analysis has already achieved 100% recognition of close-proximity vehicles, why is there a need to introduce artificial intelligence and machine learning algorithms in Section 6? Given the high computational cost of AI and machine learning algorithms, is it necessary to use these algorithms for close-proximity vehicle recognition?
7. 7. After successful close-proximity recognition between vehicles and bicycles, could an alert sound be issued? This is considering that regulations stipulate a minimum distance of 1.5 meters between bicycles and cars for the safety of cyclists.

Author Response

We would like to thank the reviewers for their comments and suggestions.

Comment 1. Recent research literature should be added, as this is a manuscript submitted in 2024, yet there are hardly any references from 2021 to 2024 in the paper.

Reply 1: Thank you for highlighting the need for a more contemporary perspective in our literature review. In response to your valuable suggestion, we have expanded on our literature review and included citations of an additional 17 recent articles of relevant interest.

Comment 2: The contributions should be summarized again, given that this is a research paper.

Reply 2: The contributions of our paper have been summarized in the Introduction and again in the Conclusion.

Comment 3: The font size of the axes in the figures should be increased to improve readability. Additionally, it is rare for studies to place the figure title on the top axis.

Reply 3: We have now increased the font size of the axes labels to improve their readability. Regarding the placement of figure titles, we have maintained the figure captions adjacent to the figure numbers for standard reference. For a few graphs, we have included a brief title at the top as an aid for quick referencing. 

Comment 4: Why is there a need to add a shaded background to equations (2), (4), and (5)?

Reply 4: The shaded background has been removed. Thanks for the feedback.

Comment 5: Why is there a need to consider vehicle speed? It seems that regulations only specify a minimum distance of 1.5 meters between bicycles and cars, without restricting their relative speed.

Reply 5: Certain countries take not only passing distance, but vehicle speed into account. For example, in France, the passing distance allowed is 1m on roads with ≤50km/h speed limit, and 1.5m on roads with >50km/h speed limit. In addition, the speed of cars impacts the danger when overtaking – cars going at a faster speed are more dangerous than those going at a slower speed.

Comment 6: If clustering analysis has already achieved 100% recognition of close-proximity vehicles, why is there a need to introduce artificial intelligence and machine learning algorithms in Section 6? Given the high computational cost of AI and machine learning algorithms, is it necessary to use these algorithms for close-proximity vehicle recognition?

Reply 6: The clustering analysis that we used is a machine learning algorithm that is used to identify close vehicle passes by identifying clusters of data points measured by our distance sensor. In a dynamic, outdoor environment, distance sensors can generate a lot of spurious data points which may impact the performance of the clustering algorithm, by extension reducing its reliability. As such, we use an AI camera to reduce the amount of spurious data generated – our distance sensors only begin measuring when the AI camera sees a vehicle.

Comment 7: After successful close-proximity recognition between vehicles and bicycles, could an alert sound be issued? This is considering that regulations stipulate a minimum distance of 1.5 meters between bicycles and cars for the safety of cyclists.

Reply 7: Our setup currently is a prototype to demonstrate the possibility of automated data collection of vehicles passing distances. We wish to provide an alternative data collection method to the manual methods that many studies have been using. Moving forward, our setup once finalized would function like a dashcam in a car and provide information upon replay. We believe that it would not be very useful for an alert to be sounded as that is not the primary intention of our study and drivers will not be able to hear the alert.

Round 2

Reviewer 1 Report

The comments have been addressed satisfactorily, I thank the authors for their attention.

Reviewer 2 Report

Following the Authors' corrections and the update of the literature, I believe that the article should be published immediately.

Reviewer 3 Report

Thanks for your response