Efficient Management of Road Intersections for Automated Vehicles—The FRFP System Applied to the Various Types of Intersections and Roundabouts
Round 1
Reviewer 1 Report
The paper introduces a method for the management of road intersections for automated vehicles. The paper states the problem and explains adequately the solution proposed. The first sections (1 and 2) clearly introduces the state of art providing a detailed description of the problem addressed and the technical concepts associated with the proposed solutions.
The overall approach of the paper is good and may be considered for publication. Nevertheless, some improvements may be made. For instance, in section 3, the authors must clarify the characteristics of the evaluation done. Maybe including a table with that information may be useful.
Section 2, related to the State of the Art, includes only information about other works. It will be very interesting to introduce also some information about the use of technology and simulation for solving problems related to autonomous vehicles. For example, a lot of work on autonomous vehicles is performed in virtual reality and a lot of studies concerning autonomous vehicles are performed in graphics simulation.
Could you please justify the potential utility of your finding in the study or in other words are you planning to apply them to real-world scenarios?
Finally, this is a list of references to be considered:
--You or me? Personality traits predict sacrificial decisions in a VR-simulated accident situation
--User interface considerations to prevent self-driving carsickness
--Passenger Anxiety when Seated in a Virtual Reality Self-Driving Car
--Driving in the Matrix: Can Virtual Worlds Replace Human-Generated Annotations for Real World Tasks? Virtual to real reinforcement learning for autonomous driving
Author Response
Response to Reviewer 1 Comments
Point 1: Could you please justify the potential utility of your finding in the study or in other words are you planning to apply them to real-world scenarios?
Response 1: Of course, the intention is to use it in real zero collision systems. The implemented system reduces the crossing times for different types of intersections.
Point 2: You or me? Personality traits predict sacrificial decisions in a VR-simulated accident situation
Response 2: Only decisions in favour of road safety are envisaged in order to optimise crossing times and therefore road congestion.
Point 3: User interface considerations to prevent self-driving carsickness
Response 3: The System foresees a management of the intersections that, applying the algorithm already at great distances, typically foresees a minimum variation of speed.
Point 4: Passenger Anxiety when Seated in a Virtual Reality Self-Driving Car
Response 4: No, we haven't analysed the passenger's possible anxiety in automatic driving. Surely in a first stage of transition it is to be seasoned as possible.
Point 5: Driving in the Matrix: Can Virtual Worlds Replace Human-Generated Annotations for Real World Tasks? Virtual to real reinforcement learning for autonomous driving
Response 5: Surely automatic guidance opens up a new world, learning systems will strengthen the optimization of automatic guidance in general.
Reviewer 2 Report
The paper proposes a novel method to manage vehicles' crossings and compare the performances of the method with those of the methods available in the literature.
This reviewer has the following comments.
1) The method requires a full communication between all the vehicles, which implies that all the vehicles must be endowed with a communicating equipment and that some central decision makes the computations and communicates the priorities and the speed variations to all the vehicles. This procedure seems unrealistic in real-time.
2) The performances of the method are comparable with those of the FCFS procedure. As a consequence, there seem to be no reason to implement the proposed FRFP procedure.
3) Figures are not very clear: what does the horizontal axis represent?
Author Response
Point 1: The method requires a full communication between all the vehicles, which implies that all the vehicles must be endowed with a communicating equipment and that some central decision makes the computations and communicates the priorities and the speed variations to all the vehicles. This procedure seems unrealistic in real-time.
Response 1: The system studied can be applied by means of centralized systems but, above all, by means of distributed V2V communication and therefore realistically applicable in a short time without the enormous disadvantage of installing centralized intersection management systems.
Point 2: The performances of the method are comparable with those of the FCFS procedure. As a consequence, there seem to be no reason to implement the proposed FRFP procedure.
Response 2: The services depend very much on the congestion conditions and the type of intersection. As can be seen from the various comparisons it is true that in some cases the reduction in both crossing time and consumption and emissions is reduced to a few percentage points, but it is also true that under high conditions we reduce crossing time by about 22% and emissions by 7.4%.
Point 3: Figures are not very clear: what does the horizontal axis represent?
Response 3: Thank you for the important suggestion, I will improve the indications of the graphs, the horizontal axis indicates the time
Reviewer 3 Report
The paper deals with management of road intersections for automated vehicles, assuming no traffic lights or priority rule at the intersection.
An analytical method is given based on the FRFP (first to reach the end of the intersection first to pass) method, assuming constant accelerations for the vehicle. In solving the intersection control problem, it is assumed that only one vehicle can stay inside the intersection at a given time. The proposed method deals with congestion avoidance as well.
The paper is well structured, but contains some mistakes in the statements (for example, the first sentence of the introduction). Some other remarks:
The references should be completed, lots of them miss to name the journal or conference published in. The equations should be numbered, especially as in the text it is cited with number (page 3). The proposed method is not clearly introduced, the calculations are often based on the same formula with different notations, and the block diagrams showing the operation of the algorithms are not explained well enough. Please clarify.Also, authors miss to address some important aspects related to intersection control. As a safety critical system affected by possible sensor faults, communication delays and data loss, robustness is very important in the design and should be considered. Simulation results carried out in SUMO show the effectiveness of the proposed method compared to conventional traffic light control or rule based methods. However, authors should point out the advantages of the proposed method compared to other solutions in the literature (artificial intelligence and game theory based methods).
Author Response
Point 1: The references should be completed, lots of them miss to name the journal or conference published in.
Response 1: Thanks for the suggestion, I'll improve the references
Point 2: The equations should be numbered, especially as in the text it is cited with number (page 3).
Response 2: Yes thank you, in fact there is a numerical reference, but not necessary, to the formula that follows the sentence. I will update the document.
Point 3: The proposed method is not clearly introduced, the calculations are often based on the same formula with different notations, and the block diagrams showing the operation of the algorithms are not explained well enough. Please clarify
Response 3: The concept is very simple and is based on the link between space, time and speed. The real novelty is the priority calculation that refers to the vehicle that has more potential to arrive first at the end of the intersection. The diagram, even if simplified, represents only the calculation methodology according to the type of motion (accelerated or constant speed).
Point 4: Also, authors miss to address some important aspects related to intersection control. As a safety critical system affected by possible sensor faults, communication delays and data loss, robustness is very important in the design and should be considered
Response 4: The work is based on the most efficient algorithm in terms of crossing times, consumption, emissions and road congestion. The problem of communication is not addressed in any way regardless of the algorithm used. Clearly, an application design phase, which is not part of the work carried out, must necessarily include robust systems for the management of anomalies of any kind.
Point 5: Simulation results carried out in SUMO show the effectiveness of the proposed method compared to conventional traffic light control or rule based methods. However, authors should point out the advantages of the proposed method compared to other solutions in the literature (artificial intelligence and game theory based methods).
Response 5: As indicated in the conclusions, the proposed system not only presents clear efficiency characteristics of the parameters under examination, but also offers the considerable advantage of not requiring the installation of intersection operators based on V2V communication systems.
Round 2
Reviewer 1 Report
After reading the revised version of the paper as well as the responses made by the authors, I feel confident that this is a strong and scientifically sound paper.
Reviewer 2 Report
Authors replied to this reviewer's previous questions. Still this reviewer thinks that the full communication between all the vehicles involved in the crossing, although all other vehicles in the rest of the communication network are neglected, is a heavy requirement. However, with this assumption, conclusions seem to be correct. For this reason, although no theoretical result (such as theorems of general validity) is shown, the article deserves publication.
