Autonomous Vehicles Management in Agriculture with Bluetooth Low Energy (BLE) and Passive Radio Frequency Identification (RFID) for Obstacle Avoidance

Round 1

Reviewer 1 Report

The paper presents an analysis of the problem of obstacle avoidance and autonomous vehicles management in Agriculture. The analysis concerns only objects equipped with some markers readable for other mobile objects in the system. I highlight that worker without these markers are invisible for the autonomous vehicles. This is the main disadvantage of the presented system.

Some procedures for analyzing the operation and security of systemsof (HTA, SPOFs ,FMandEA) are applied first. These formal considerations are rather simple and consider only the distance between the active objects as a parameter of the interaction. This analysis is not exhausting. It ignores, for example, the case when the objects are close but moving on the parallel trajectories.

In Figure 6 the flowchart of the proposed safety system is presented. According to this flowchart, again the only parameter taken into account by the safety system is the distance threshold. It is two-stage collision risk assessment system based on the distance from the obstacle. The simplicity of this system probably guarantees its reliability. It would be interesting, however, how the practical values of the alert and critical distances are determined, depending on the speed of moving vehicle. Why the relative trajectories of the objects are not estimated? In the paper the AoA is mentioned, and the reception of the signals is executed by two receptors located at a distance on the vehicle. Why these data are not taken into consideration in the flow chart of the system? Why the analysis of previous distance estimations are disregarded? the acronym of the project suggests the possibility of locating objects in relation to the mesh of points. Is the Figure 9 drawn in the real time? Is this the “on line” information for the operator?

Author Response

We thank the reviewer for the insightful comments.  Responses from authors follow below. 

Point 1: The paper presents an analysis of the problem of obstacle avoidance and autonomous vehicles management in Agriculture. The analysis concerns only objects equipped with some markers readable for other mobile objects in the system. I highlight that worker without these markers are invisible for the autonomous vehicles. This is the main disadvantage of the presented system.

Response 1 (IMPLEMENTED) - INAIL's (National Institute for Insurance against Accidents at Work, funder of the project) additional goal was, in fact, to exploit RFIDs placed on PPEs for this purpose. The project includes RFID scanners (gates) at the entrance of the working areas, preventing workers from accessing without PPEs. Authors better specified this concept in lines 234-237 on manuscript V2.

Point 2: Some procedures for analyzing the operation and security of systemsof (HTA, SPOFs ,FMandEA) are applied first. These formal considerations are rather simple and consider only the distance between the active objects as a parameter of the interaction. This analysis is not exhausting. It ignores, for example, the case when the objects are close but moving on the parallel trajectories.

Response 2 (IMPLEMENTED) - Table 6 is only meant to be a generic representation of the main failure modes in agricultural activities involving both machines and workers on foot altogether. Although, a deep FMEA analysis of those accident scenarios had already been performed by authors and summarised in that table: mechanical harvest of shell fruits (row no.1), in fact, does implicitly take into account parallel trajectories among tractors/self-propelled machines and workers ("wrong distance with the equipment while working"). In order to make it clear that row no.1 applies to any activity involving parallel trajectories and not only to shell fruit harvest, authors improved the table to make it clearer.

Point 3: In Figure 6 the flowchart of the proposed safety system is presented. According to this flowchart, again the only parameter considered by the safety system is the distance threshold. It is two-stage collision risk assessment system based on the distance from the obstacle. The simplicity of this system probably guarantees its reliability. It would be interesting, however, how the practical values of the alert and critical distances are determined, depending on the speed of moving vehicle.

Response 3  (PARTIALLY IMPLEMENTED) - Thresholds should be determined based on risk assessment: indeed, speed of the vehicles(s) is a critical aspect, but same goes for number of workers, type of activity and driver's/operator's visibility. Since we imply this in the system's concept, we have better specified this in the text after figure 6.

Point 4: Why the relative trajectories of the objects are not estimated? In the paper the AoA is mentioned, and the reception of the signals is executed by two receptors located at a distance on the vehicle. Why these data are not taken into consideration in the flow chart of the system? Why the analysis of previous distance estimations are disregarded? the acronym of the project suggests the possibility of locating objects in relation to the mesh of points.

Response 4  (NOT IMPLEMENTED) - Trajectory estimation could be possible for well-defined movements only, like vehicles. Despite exact positions can be determined by the system, its error margin would cumulate when detecting workers trajectories. In case of repeated movements by workers, this might also lead to false alarms since the system wouldn't be able to adapt to them and at the same time it might lead to accidents if an algorithm recognises them as repeated while they aren't. Similar issues are experienced by SAE level 5 autonomous car (fully automated vehicles) and that's one of the reasons why nowadays it is considered a merely theoretical level and not yet reachable. Data from previously identified obstacles could of course be stored in the back-end server (like system does for near-misses), but it would be useful for fixed obstacles only or, at most, for a certain amount of time after which data will become outdated.

Point 5: Is the Figure 9 drawn in the real time? Is this the “on line” information for the operator?

Response 5 (PARTIALLY IMPLEMENTED) – Figure 9 is drawn from real time data but its information would be reported to the operators only if distance is below alert/danger thresholds, and it might even result in emergency stops alarm on the machine. Reporting such live information inside the cab might require an additional LCD on most tractor models, hence it has been discarded in project design.

Reviewer 2 Report

Authors are invited to improve the title and possibly to avoid the use of acronyms in the title.

In any case it is advisable to avoid the RFID acronym without the complete wording (as it is for the BLE acronym). A suggestion is “Autonomous vehicles management in agriculture to avoid obstacles”.

 

English language has to be improved in order to better comprehend the subject

Abstract

PPE, personal protective equipment, check and update the acronym across the manuscript.

 

RFID, radio frequency identification, even if the acronym is firstly introduced in the abstract, it is advisable to repeat the wording also in line 59 in the manuscript.

 

Introduction

Lines 37 to 41, improve the sentence to comprehend the point.

Materials and methods

Table 6, the readability and the quality of the table need improvements

Line 151 to 153, please improve the sentence.

Line 171, how many tractors have been tested. Based on Figure 1, one tractor was tested.

Lines 176 to 181, please improve the sentence, unclear as written.

Results

Line 202, a verb is missed in the sentence.

Line 216, sides or sizes?

Lines 220 to 227, please improve the paragraph, unclear meaning as written.

Lines 227 to 230, additional explanation is deemed necessary to comprehend Figure 8.

Lines 234 to 238, additional details on the preliminary tests carried out can help to better comprehend Figure 9.

Figure 9 caption, it is advisable to avoid wording “example of filed test…”

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments to the Author

 

This paper describes the SMARTGRID project which integrates a network of BLE devices and RFID tags to create a wireless safety network infrastructure. The network of sensors is designed to identify obstacles, workers, nearby vehicles and check if the right PPEs are in use. It is an interesting topic and the paper studies the concept clearly. However, there are several points need to be addressed to improve the quality of the manuscript.

 

Suggestions to improve the quality of the paper are provided below:

 

1)     The novelty of this work needs to be highlighted more clearly by explaining how the proposed approach differs or extends upon existing works. Currently, the authors have only listed out the relevant works in Table 2 but does not go into details about the technologies adopted by those works.

 

2)     The authors should also provide a stronger justification regarding the technology that they have adopted. Why did they decide to go with BLE and RFID technologies? Why are these technologies suitable for this particular use case?

 

3)     What are the other successful application areas of these technologies in the literature? Please go through the following studies and include them as part of the literature review.

 

Localisation for building emergency management using BLE technology

Filippoupolitis, A., Oliff, W., & Loukas, G. (2016, December). Bluetooth low energy based occupancy detection for emergency management. In 2016 15th international conference on ubiquitous computing and communications and 2016 International Symposium on Cyberspace and Security (IUCC-CSS) (pp. 31-38). IEEE.

 

Filippoupolitis, A., Oliff, W. and Loukas, G., 2016, October. Occupancy detection for building emergency management using BLE beacons. In International Symposium on Computer and Information Sciences (pp. 233-240). Springer, Cham.

 

Localisation for occupancy tracking in office spaces using BLE technology

Tekler ZD, Low R, Blessing L. An alternative approach to monitor occupancy using bluetooth low energy technology in an office environment. In Journal of Physics: Conference Series 2019 Nov 1 (Vol. 1343, No. 1, p. 012116). IOP Publishing.

 

            Mapping and localisation using RFID technology

Hahnel, Dirk, et al. "Mapping and localization with RFID technology." IEEE International Conference on Robotics and Automation, 2004. Proceedings. ICRA'04. 2004. Vol. 1. IEEE, 2004.

 

Occupant behaviour monitoring using BLE technology

Tekler, Z.D., Low, R., Gunay, B., Andersen, R.K. and Blessing, L., 2020. A scalable Bluetooth Low Energy approach to identify occupancy patterns and profiles in office spaces. Building and Environment, 171, p.106681.

 

            Building performance evaluation using RFID technology

Li, N. and Becerik-Gerber, B., 2011. Performance-based evaluation of RFID-based indoor location sensing solutions for the built environment. Advanced Engineering Informatics, 25(3), pp.535-546.

 

4)     In section 2.2, the authors identified three well known agricultural activities for performing the Hierarchical Task Analysis. However, it is unclear why are these activities “well known”. Are there many accidents linked to these activities? Furthermore, a more detailed description for each activity is necessary to provide the readers, who are unfamiliar with these activities, more context to understand the content in Table 6.

 

5)     It is unclear how BLE beacons/technology is used within the proposed system.

 

6)     The authors mentioned that “The performance of the system can also be affected by the number of RFID devices on it and by their positioning.”; however, there is no quantitative results to show how the performance is affected.

 

7)     The authors should spend some time to discuss about the practicality of the experimental setup and the system’s deployment in real-world conditions as the Discussion section is lacking. This includes:

a.     Does the posture of the workers (i.e., standing, squatting, crawling) have an impact on the systems’ localisation accuracy? If the answer is yes, please consider what can be done to minimise its impact (e.g., RFID tag placement, etc)?

b.     Based on the experimental description and Figure 7, it seems like stationary poles are used to simulate the workers. If that is an accurate description of the experiment conducted, it is then necessary for the authors to also consider more realistic conditions such as non-stationary targets and its effect on the localisation accuracy.

c.      Did the authors consider the system’s latency between detecting the workers’ location and the notification being sent out? I believe this has to be considered when determining the danger thresholds.

d.     How does natural obstacles affect the system’s localisation accuracy?

 

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

I did not receive a comprehensive answer to the questions that I asked in the opinion. When analyzing the introduced changes, I understand that the authors want to keep the system as simple as possible, believing that this will result in its reliability. I also understand that by this the issue is not similar to that of autonomous vehicle control in road traffic. Such comments should be included in the introduction. The proposed system will probably be safe, according to the authors, so the work may be published, but I believe that there is no scientific news in the adopted solution. It should be a technical note or a case study.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Title - Authors are invited to consider modifying the manuscript title as “Autonomous vehicles management in agriculture with Bluetooth Low energy (BLE) and passive Radio Frequency Identification (RFID) for obstacle avoidance”

Line 147 – it is advisable to revise the paragraph deleting the word pre-pruning as follow “ … since pre and post-pruning operations are necessary, vehicles need to be preceded by workers on foot and followed by other …” 

Line 181 – it is advisable to replace “kind of warning” as “warnings”.

Lines 271 to 274 – improve the English in the paragraph

Line 304 to 305 – delete the double “that”. Moreover, posture is it referred to “worker posture”? Consider then to replace “at least some of them” with “an acceptable number of RFID” 

Line 316 – consider replacing “to increase to test more realistic conditions like simulating the presence of workers on foot” with “to test conditions more representative of the presence of workers on foot”

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Thank you for addressing my concerns. I recommend this manuscript in its present form.

Author Response

Authors thank the reviewer for the feedbacks.