Interactive System for Package Delivery in Pedestrian Areas Using a Self-Developed Fleet of Autonomous Vehicles

Round 1

Reviewer 1 Report

This describes a vehicle design scheme that can independently plan the route to realize parcel delivery. The design scheme has fully considered the personalized needs of users and allows users to participate in the decision-making process. Users can submit instructions through the interface on the vehicle or through the application on the mobile phone. After a detailed introduction of the vehicle's implementation scheme and some detailed features, the author did a preliminary test in a strictly restricted area and proved that the system can work properly.

1. The author claims that the biggest innovation of this paper is the addition of interaction with users. However, in the field of logistics services, how to achieve "Last kilometer" distribution through vehicles that can automatically plan routes is not only a hot research topic, but also a commercial direction that can be implemented. There are some products on the market including driving slowly on sidewalks and driving at high speeds on roads. For example, Nuro's R2 unmanned delivery vehicle, which has been commercially licensed by the government, and Amazon's Scout, Robby Technologies, Kiwibot, Eliport, TeleRetail, BoxBot, Starship Technologies and others have similar products, some of which also offer mobile applications to customers, In this way, customers can choose their own delivery time and other parameters, as well as check their current location in real time. Therefore, what are the unique differences between the author's innovations and these companies?

2. In Question 1, I have listed the names of several companies that have provided similar products. The author may list a table in the background section to compare the performance of several key parameters with products on the market.

3. Grammar problems may exist in some places. For example, line 156: 'These methods do not consider peoples demands and wished and only optimize travelling time or distance'.

4. The main research focus of this paper needs to be further highlighted.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The authors described the design and implementation of system integration of an autonomous vehicle fleet for package delivery. The authors have conducted extensive field tests to demonstrate the effectiveness of the decribed system.

The novel contribution of the presented work is not clear. The authors only employed well investigated methods/algorithms and path planning and navigation algorithms provided in ROS for implementation of the system.

In Section 2, literature review is not thorough, especially on localization and path planning. The authors barely mentioned standard approaches for localization and path planning developed in mobile robotics without putting them under the context of autonomous vehicle fleets. The authors are suggested that they focus on comparison of design and performance of existing deployed autonomous vehicle fleets system integration in details. For the mission planning/logistic planning aspects, algorithms used in Uber system or food delivery systems may be surveyed.

Figure 3, Figure 4, and Figure 23 are splitted in 2 pages. The authors are suggested that they keep all subfigures in the same page.

The novelty of the proposed mission planning subsystem is not highlighted in the manuscript. The design and the algorithms adopted in this subsystem are widely adopted. It also seems that the description of the proposed mission planning subsystem do not focus on an autonomous vehicle fleet, but on single vehicle.

How are the missing planner and path planner integrated to produce the cost (travelling time) between pairs of nodes/stations?

In p.10, the authors mentioned that when a user makes a request inside an available slot, the two existing missions m_k and m_{k+1} should be updated to be m_k' and m_{k+1}' respectively. Why only these two missions need to be updated? how about the subsequent missions m_{k+2}, m_{k+3}, etc.?

The novelty of the transformation of an electric golf cart to an autonomous package delivery vehicle is not clear. The overall design of the system is very similar to a common commercial electric vehicle on street.

The authors are suggested that they provide more technical details on the generation of AMCL estimates. Why AMCL is adopted in this work? Have you evaluate the performance of sensor fusion based localization with other localization estimates based on LIDAR data?

The novelty of the path planning subsystem is not clear. The authors mentioned that they applied the built-in sbpl_lattice planner in ROS for path planning on modified map (with corridor restriction). The addition of corridor restriction in the map in not a new concept in path planning. It is only an example of forbidden regions added in configuration space common in path planning. Built-in navigation algorithm in ROS is also applied without any modification in the discussed system. More details on the role of onboard sensing systems, like ultrasonic sensors, LIDAR, vision systems and bumpers, in navigation and collision avoidance are expected in the manuscript.

How many autonomous vehicles were involved in testing? In line 970, the authors mentioned two vehicles were involved. But, there are three vehicles shown in Figure 32(b). Were they operated at the same time? 

In line 833, "Figure 7.10" should read "Figure 22". In lines 960 and 962, "Fig. 7.15, 7.16 and 7.17" are not shown in the manuscript.

What are the units used in the deviations mentioned in line 869, Table 6 and line 924? meters or centimeters?

Lots of spelling and grammatical mistakes are spotted in the manuscript. The authors are suggested that they conduct thorough proofreading on the manuscript.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

This paper is too long at 34 pages. The electrical design of the autonomous robot is interesting, so are the driving experiments. I think the paper must be shortened to make a specific point, rather than show all the results you have. More importantly, you mention pedestrians very often (62 times), you don't describe what algorithm is used to detect pedestrians. From what I see, they are detected using a "vision sensor". The following are some comments about writing: 

• error on line 154
• on line 960 you refer to, no-existent figures 7.15 and 7.17
• the way you cite the references in the text is incorrect. You don't say, "xyz is shown in [12]" referring to [12] as an object. Also, write [12-14] instead of [12, 13, 14]

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

The authors have addressed my comments raised in the previous around of review. The submitted manuscript presented an interesting case study of deployment of an autonomous fleet management system for outdoor delivery. The authors have conducted extensive experimental investigation in the field to show the effectiveness of the proposed system. The manuscript is recommended for publication in the Electronics journal after careful proofreading for eliminating spelling and grammatical errors.

Author Response

Dear madam or sir,

thank you for your suggestions. The paper has now undergone the English language editing by MDPI.

Best regards,

Mihai Kocsis

Reviewer 3 Report

The paper/report is still too long at 30+ pages. There seem to be two topics, high-level path planning and the technical development of the vehicle. The paper should be shortened and focused on one theme. It is better if it is split into two separate papers.

Author Response

Dear madam or sir,

thank you for your review and suggestions. We discussed with the editor and decided to keep it like this because it is a complete article and offers a global vision in itself. If we divide it in to two separate articles it would lose much of its meaning. Also the other reviewers share this opinion.

After final proofreading for spelling and grammar mistakes we will submit the manuscript.

Thank you and best regards,

Mihai Kocsis