Reviews and Comments on Paper 561

Paper information

Paper #561: Luz Abril Torres-Mendez and Gregory Dudek. Statistics of Visual and Partial Depth Data for Mobile Robot Environment Modeling

Abstract: In mobile robotics, the inference of the 3D layout of large-scale indoor environments is a critical problem for achieving exploration and navigation tasks. This article presents a framework for building a 3D model of an indoor environment from partial data using a mobile robot. The modeling of a large-scale environment involves the acquisition of a huge amount of range data to extract the geometry of the scene. This task is physically demanding and time consuming for many real systems. Our approach overcomes this problem by allowing a robot to rapidly collect a set of intensity images and a small amount of range information. The method integrates and analyzes the statistical relationships between the visual data and the limited available depth on terms of small patches and is capable of recovering complete dense range maps. Then, all these augmented range maps are integrated to build a 3D model of the environment. Experiments on real-world data are given under different configurations to illustrate the suitability of our approach.
(file)

Summary of received reviews and comments

Reviews superseded by other reviews are shown in the grey color in the table.

	confidence	score
Review 1	3	2
Review 2	4	3
Review 3	4	3

Reviews and Comments

Review 1

PC member:
Reviewer:
Overall rating:	2 (accept: I will argue for this paper)
Confidence:	3
Relevance: Is this paper relevant for this conference?	3 (strong accept)
Soundness: Is this paper technically sound and complete?	2 (accept (I will argue for this paper))
Are the claims sufficiently supported by experimental/theoretical results?	1 (weak accept (vote accept but don't mind rejecting))
Significance: Are the results/ideas interesting for other AI researchers?	3 (strong accept)
Originality: Are the results or ideas novel and previously unknown?	1 (weak accept (vote accept but don't mind rejecting))
Readability: Is the paper well-organized and easy to understand?	2 (accept (I will argue for this paper))
Language: Is the paper written in correct English and style?	2 (accept (I will argue for this paper))
Format: Is the paper correctly and consistently formatted?	3 (strong accept)
Review:	CONTRIBUTION OF THE PAPER: The paper describes a method for constructing dense range panoramas by fusing sparse range data (e.g. laser scanner) and intensity images. The intensity and range images are registered given a known camera/laser scanner geometry, and a dense range image is estimates using a MRF that relates image intensities to range correlations and values. POSITIVE ASPECTS: The paper is very well written (apart from some notational issues - see below), lays out the problem well, discusses an interesting model of the sensor fusion and presents some good results. NEGATIVE ASPECTS: The primary issue I have with the paper is the experimental validation - just one image is not really enough to get a good idea. The authors should evaluate their error for different levels of missing data, and plotting that in a graph. also, it would be good to compare against a simple heuristic method to see how much the MRF is adding - say use a mean shift to segment the image and assign ranges from the range data directly to regions? Second issue I have is the backgrouund - it seems that MRFs have been used before to approach this problem, but you have no references about that - e.g. Segmentation based on fusion of range and intensity images using robust trimmed methods IS Chang, RH Park - PATTERN RECOGN, 2001 There are some notational/presentation issues (fairly minor i think) that made the paper a little hard to read - see below. CHANGES TO IMPROVE THE PAPER: notational issues: there seems to be some loose use of the letter 'v' and the greek letter 'nu' that seem to refer to the same thing? I think introducing the coordinate systems in a graph at the start of 4.2 and defining and labeling all your variables would be very helpful. I think the presentation of figure 3 could also be improved - maybe add a bit of shading somewhere to make the perspective work? Finally, figure 5 needs a much better caption and more clarity/relationship with the text. I suggest introducing this figure earlier and relating to it in the text more clearly. FURTHER COMMENTS: ITEMS BELOW ARE JUSTIFICATION OF THE SCORES IF NEGATIVE: (1) IS THIS PAPER RELEVANT FOR THIS CONFERENCE? (2) IS THIS PAPER TECHNICALLY SOUND AND COMPLETE? (3) ARE THE CLAIMS SUFFICIENTLY SUPPORTED BY EXPERIMENTAL OR THEORETICAL RESULTS? (4) ARE THE RESULTS/IDEAS INTERESTING FOR OTHER AI RESEARCHERS? (5) ARE THE RESULTS OR IDEAS NOVEL AND PREVIOUSLY UNKNOWN? (6) IS THE PAPER WELL-ORGANIZED AND EASY TO UNDERSTAND? (7) IS THE PAPER WRITTEN IN CORRECT ENGLISH AND STYLE? (8) IS THE PAPER CORRECTLY AND CONSISTENTLY FORMATTED?
PC only:
Time:	Jul 13, 12:05

Review 2

PC member:
Reviewer:
Overall rating:	3 (strong accept)
Confidence:	4
Relevance: Is this paper relevant for this conference?	3 (strong accept)
Soundness: Is this paper technically sound and complete?	3 (strong accept)
Are the claims sufficiently supported by experimental/theoretical results?	3 (strong accept)
Significance: Are the results/ideas interesting for other AI researchers?	2 (accept (I will argue for this paper))
Originality: Are the results or ideas novel and previously unknown?	2 (accept (I will argue for this paper))
Readability: Is the paper well-organized and easy to understand?	3 (strong accept)
Language: Is the paper written in correct English and style?	3 (strong accept)
Format: Is the paper correctly and consistently formatted?	3 (strong accept)
Review:	CONTRIBUTION OF THE PAPER: The idea of using intensity information to guide range data restoration is quite interesting. The approach presented builds on many known techniques in a novel manner. POSITIVE ASPECTS: In particular the reconstruction during belief propagation guided by intensity information is innovative and based on good insight into the use of the MRF. The method for recovering missing range data in a system that operates on panoramas of registered range and intensity data in the lab is well demonstrated. NEGATIVE ASPECTS: None in particular. CHANGES TO IMPROVE THE PAPER: None. FURTHER COMMENTS: ITEMS BELOW ARE JUSTIFICATION OF THE SCORES IF NEGATIVE: (1) IS THIS PAPER RELEVANT FOR THIS CONFERENCE? Yes, it is a strong robotics paper. (2) IS THIS PAPER TECHNICALLY SOUND AND COMPLETE? Yes, it is clear and well argued. (3) ARE THE CLAIMS SUFFICIENTLY SUPPORTED BY EXPERIMENTAL OR THEORETICAL RESULTS? Yes, the experiments are on real robots and are convincing. (4) ARE THE RESULTS/IDEAS INTERESTING FOR OTHER AI RESEARCHERS? Yes, this will be of general interest. (5) ARE THE RESULTS OR IDEAS NOVEL AND PREVIOUSLY UNKNOWN? No. (6) IS THE PAPER WELL-ORGANIZED AND EASY TO UNDERSTAND? Yes. (7) IS THE PAPER WRITTEN IN CORRECT ENGLISH AND STYLE? Yes. (8) IS THE PAPER CORRECTLY AND CONSISTENTLY FORMATTED? Well organized and clearly presented.
PC only:
Time:	Jul 15, 22:33

Review 3

PC member:
Reviewer:
Overall rating:	3 (strong accept)
Confidence:	4
Relevance: Is this paper relevant for this conference?	3 (strong accept)
Soundness: Is this paper technically sound and complete?	3 (strong accept)
Are the claims sufficiently supported by experimental/theoretical results?	2 (accept (I will argue for this paper))
Significance: Are the results/ideas interesting for other AI researchers?	1 (weak accept (vote accept but don't mind rejecting))
Originality: Are the results or ideas novel and previously unknown?	2 (accept (I will argue for this paper))
Readability: Is the paper well-organized and easy to understand?	2 (accept (I will argue for this paper))
Language: Is the paper written in correct English and style?	3 (strong accept)
Format: Is the paper correctly and consistently formatted?	3 (strong accept)
Review:	The paper presents a statistical method for synthsizing complete cylindrical range images from partial range data and complete intensity images. The goal is to speed up scene data acquisition by partial sampling and synthesis. Markov random field models of the local correlated intensity and range structure are used. This was the clearest and most 'professional' paper in my set. The work is fine as it stands, but several improvements are suggested: 1) analyze the reconstruction failures rather than simply quoting the MAP error - this might suggest areas/techniques of improvement. 2) You might compare your statistical reconstruction approach to the sampling approach of Efros (2D) and Breckon (3D), or exploiting surface continuity (Fisher's group). 3) Eqn's 4, 5, 11 seem to have typo's in them.
PC only:
Time:	Jul 17, 16:18