SoC-VRP: A Deep-Reinforcement-Learning-Based Vehicle Route Planning Mechanism for Service-Oriented Cooperative ITS

Round 1

Reviewer 1 Report

The article uses mathematical methods and modern simulation approaches. Correctly, it also takes into account different types of vehicles and their different levels of density in the process. It also takes into account the various travel restrictions in the analyses and the importance of service-oriented collaborations. Creates a sequential decision route planning method based on changes in vehicle density. In this regard, it would have been correct to say, for example, that in this case, in terms of network processes, the system state characteristics are the vehicle densities occurring on the sectors. The authors present in detail how, with the support of new information technology, transportation becomes an intelligent system. It is also worth mentioning, in regarding the article that it provides a detailed overview of the related modern algorithm theory areas and provides a summary of this as well. The authors rightly point out that the use of Markov decision processes (MDP) is extremely important for the analysis of traffic processes. Reinforcement learning is used, in which learning is combined. Focusing on  route planning, all roads have been simplified to one lane and reward/goal functions are combined. The well-proven, validated SUMO (Simulation of Urban Mobility) extended simulator software is used for microscopic modeling. Since the work presented in the article serves the optimization of road traffic on a wider scale:

I recommend highlighting in the summary that, in addition to the above, this work also helps the economical, energy-saving operation of road vehicles - which, in the case of non-electric vehicles, also affects the reduction of environmental pollution, see e.g. the following literatures - which I also recommend referring to:

Lakatos István, An integrated analysis of processes concerning traffic and vehicle dynamics, development of laboratory applying real traffic scenarios. (2018) INTERNATIONAL JOURNAL OF HEAVY VEHICLE SYSTEMS 1744-232X 1741-5152, 3427771

Lakatos István. Diagnostic measurement for the effective performance of motor vehicles. (2013) ACTA POLYTECHNICA HUNGARICA 1785-8860 10 3 239-249, 2168984

The English text is easy to understand. At the same time, it is worth reading the article once more so that it is clear in all respects.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

This paper describes a Deep-Reinforcement-Learning-based prioritized route planning approach for cooperative Intelligent Transportation systems.

The entire approach is well described. The testing of the approach is thorough enough. Based on the description of the approach and the performed tests, the entire approach seems to be sound.

The structure of the paper is good. The related work, which is in Section 1, should be moved to a new separate Section 2. Nevertheless, if related work is required to be in Section 1 by the journal template, ignore this comment.

The figures are appropriate, but some in-figure captions are to small to be legible (e.g., in Fig. 9).

The references seem to be relevant and up to date.

The English is good, the amount of typos and errors is quite low. Still, proof-reading by a grammar-skilled native speaker could bring some improvement.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 3 Report

The paper is fine, the structure is clear, the figures are easy to understand. The aim of the research is clear. The results are well introduced. I like the paper.

Thank you for the clear focus and explanation.

Author Response

Please see the attachment.

Author Response File: Author Response.docx