Optimization of Wheelchair-Mounted Robotic Arms’ Base Placement by Fusing Occupied Grid Map and Inverse Reachability Map
Round 1
Reviewer 1 Report
This article describes a system to increase the autonomy of a wheelchair equipped with a robotic arm. This objective is achieved by optimizing the base placement of the wheelchair according to the object to reach, fusing data coming from the occupied grid map and the inverse reachability map.
I like the solution, and how it is implemented and explained in the paper. It may certainly reduce the amount of time devoted by the user to control the robotic system.
But I have one main concern: experiments do not cover, in my oppinion, a broad enough set of situations and users to be really representative.
On one hand, the experiments seems to involve telecontrol. No images are presented in which an user is seated in the wheelchair. Hence, it is difficult to infer if the solution is really helping or not a real wheelchair user. The text say the person uses the touchscreen (mounted on the robot) but the wheelchair seems to be moving around without a person seated on it.
On the other hand, the test environment is controlled, or it seems to be controlled. How would the system react in a dynamic real environment, where people walk around, objects can change position during the day, etc etc? The authors discuss in the conclusion that the system may be slower to work in such situations, and that further work could focus on increasing speed... I would like a deeper analysis of the feasibility of it: robotic solutions working in daily life scenarios are becoming an usual thing, this proposal should at least guarantee it will be able to do so.
The authors should include in the paper not only tests with people seated on the wheelchair, but also with people in an user cohort that matches as close as possible the most probable users of a wheelchair with a robotic arm. They should present (anonimous) user data in order to validate, for instance, if this solution is valid for an elderly person.
Finally, the text should be reviewed to correct some typos and errors.
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
The authors proposed the optimization method for base placement of wheelchair-mounted robotic arm by fusing occupied grid map (OGM) and inverse reachablility map (IRM). This paper presents the algorithm, the actual robot and experiment in two different tasks.
The paper is organized well. The reviewer has some questions:
1. In abstract, the proposed method improve the performance in both operation time and task completion time by comparing to the joystick control (line 22). For my opinion, the performance of proposed method should also compare with other algorithm. Why the authors not mention EL-E algorithm ?.
2. Figure 4 shows Cartographer flowchart. It is unclear that your work use only camera or not. Several sensors used in this work was mentioned in Figure 3 and line 185-189.
3. Figure 7, how values in Inverse reachability map calculated. What is the criteria to define as high, reachability and low.
4. The result shows only start position and final position. It could be better if the authors present the based position results that provides by the proposed method and other methods.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
Dear authors,
Thanks for your answers. I think this is a paper that points towards a very good direction, it's well written and presents interesting results. I would accept it without doubts if it would focus, for instance, on a robotic solution for autonomous picking, in a controlled environment.
However, the system you are proposing includes a robotic arm mounted on a wheelchair, not an autonomous robot that will be working without a person getting close to it. Hence, it is a key part of the design process, from the very beginning, to take into account the needs and preferences of the user in the wheelchair.
The robotic arm should not only consider just reachability, but also a lot of additional criteria in terms of safety (as you comment in your review), comfort, and acceptability. All that influences the movement of the wheelchair and the robotic arm. It also affects interfacing the touch screen. And all that criteria is not to be added a posteriori, but it should be considered throughout all the design and implementation process.
I think the paper should focus on a different solution (not a wheelchair equipped with a robotic arm), or include at least the first steps towards achieving user safety and acceptability. Even a virtual representation of the human in the wheelchair to be considered when computing the reachability space would add something in that direction (but the problem is of course bigger, as that user may move her arms, etc). I would encourage authors to provide those insights into what the real application of this system will be in order to get a more solid paper. And I definitely would like to read that new paper.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 2 Report
I accept your revised version manuscript.
Author Response
Thanks for your constructive comments.
Round 3
Reviewer 1 Report
Dear authors,
thanks for including those descriptions of previous work in the paper. They certainly provide the context I was missing, and from my point of view these few paragraphs change the paper quite a lot. I would still have prefered to have considered the person sitting on the wheelchair in the tests discussed in the paper (and I encourage you to do so in the future), but I think it can be accepted in its current form.
