Integration of Real-Time Semantic Building Map Updating with Adaptive Monte Carlo Localization (AMCL) for Robust Indoor Mobile Robot Localization

Round 1

Reviewer 1 Report

This paper proposes a modified Adaptive Monte Carlo Localization (AMCL) integrating object recognition and map updating into the traditional probabilistic localization. Generally speaking, the paper is very well written, the proposed algorithm is very novel to improve the localization accuracy. However, it is recommended that the authors modify the following details to improve the quality of the paper.

1. Heading 4.2 is easy to be ignored, it can be considered to be placed on the first line of page 8;

2. The indentation of the first line of the second-to-last line on page 15 is unnecessary, and a spacer can be used instead;

3. There are some details in the pictures that need attention, for example, the first line of picture description text in Figure 13 on page 18 is not aligned;

4. The literature review section can add content related to object recognition, point cloud processing and map update.

Author Response

Thank you for your constructive comments. Please find our answers for reviewer 1. 

Author Response File: Author Response.pdf

Reviewer 2 Report

The study proposes an AMCL-based indoor robot localization method improved in terms of accuracy and robustness, which is combined object recognition and semantic three-dimensional building map dynamical updating.

The paper is well organized. The authors introduced previous work and straightforwardly described the proposed method. Regarding objectives, it is possible to observe clarity in the definition of the hypothesis. Moreover, the scientific and social relevance of the proposal is welcome to the mobile robot community.

The authors present a methodology for the experiments and good enough results which justify the research. This kind of research is significant because mobile robots are used in a broad environment spectrum.

Author Response

Thank you for your constructive comments. Please find our answers for reviewer 2. 

Author Response File: Author Response.pdf

Reviewer 3 Report

Authors analyze the state-of-the-art in existing localization techniques, identifying the drawbacks and limitation of such ones, and they propose an interesting improved AMCL-based method that considers elements, which could change its position with certain low probability but resulting in problems for localization if such changes occur.

They evaluate the performance of the method for both global localization, including the kidnapped robot problem, and localization for tracking the position.

The experiments are suitable, since a digital twin is used for experimentation in a simulator, but results in a real environment is also presented.

The work is representative and it should be accepted with some minor improvements.

In particular, it is not clear how the map is updated when a specific obstacle is detected and projected, after acquiring the information from the rgb-d sensor. I understand that a 2D projection of the detected element is included in the metric map, but what happends is AMCL is failing just before watching such specific object, is it then introduced in the map in a wrong position? 

Please, explain better this point.

Some minor revision of the text regarding syntactic issues should be done.

For example:

In 3nd line at the beginning of the introduction, I think it should be "information of the environment". The word "of" is missing.

You say that laser range finders are expensive and it is not possible to find low cost implementations, however, several low-cost LIDAR options are available in the market. Reference 22 is obsolete in this sense.

In page 4, at the end of the first paragraph, you uses the acronym BIM, but its meaning is not previously written, I guess.

In the first paragraph of line 15 in page 6, it should be say "3d building model is updated when a robot (...)"

In section 3, Line 4, are you sure to say none of the existing is correct? Maybe saying "none of the mentioned approaches" would be more true.

Many of the references are older than 10 years. Have you found any reference addressing this issue in recent years? If not, it would be interesting to dicuss a possible justification for not finding them. Is is difficult to find them? Do not exist similar approaches?

Please, include more recent references related to the addressed issue.

Author Response

Thank you for your constructive comments. Please find our answers for reviewer 3.

Author Response File: Author Response.pdf