Ontology-Based Framework for Cooperative Learning of 3D Object Recognition
Round 1
Reviewer 1 Report
Authors have done significant work in explaining and analyzing, and explaining the idea they are proposing. However, I have some comments related to the paper as the following:
- Authors state that: “: Among the problems preventing widespread adoption of advanced service robots are 1) cost, 2) power consumption, …” Did the authors estimate how much the prices and other costs changed as a result of applying their proposed framework?
- Figure 3, Figure 4, Figure 6, Figure 7, Figure 12 should be changed to be more visible.
- The contribution of the paper should be clearly mentioned. The novelty of the work with respect to the available literature should be highlighted.
- What is the sensitivity of the proposed method? Have authors checked it?
- How the proposed method can be regarded as more effective and efficient in comparison with existing works?
Summing up, the paper should be improved according to the described comments and questions.
Author Response
Dear Reviewer,
We thank the reviewer for the comments. Please see theattachmentt for our responses.
Best Reguard,
Parkpoom C.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The article entitled "Ontology-Based Framework for Cooperative Learning of 3D Object Recognition" presents an interesting framework to allow robots to collaborate and learn together by using and aggregating knowledge explicitly defined in an ontology. The proposed approach works on a distributed architecture, the robots acquire 3D data which are transmitted to servers performing object detection to extract knowledge which will be stored and then shared between the robots.
The choices made for the different components of the framework are interesting. The main contribution of the article seems to be the creation of a new ontology to describe different objects but it is not explicitly mentioned. The proposed approach is sufficiently well described and illustrated to be replicable.
The main weakness of the article is the lack of research on object detection in 3D point clouds which are based on semantically guided processes. Object detection in 3D point clouds is a major topic of the article, therefore such research is needed. This lack of research is also present on the topic of where the article does not discuss the self-learning ontology for 3D point cloud approaches.
The experiments conducted to evaluate the approach are well detailed. These experiments allow a global evaluation of the framework but do not provide sufficient details on the quality of object recognition. It would be necessary to evaluate the approach on more data with different objects. Furthermore, it would be interesting to perform this evaluation using commonly used metrics such as the f1-score and IoU.
It would be helpful to explicitly describe the contributions of the article in the introduction.
line 157 mentions a new ontology presented in the article "we propose a new ontology, the Robotic Object Description (ROD) ontology ", however line 206 states that this ontology is an extension of the SUMO ontology "we propose the Robot Object Description (ROD) ontology as an extension of the SUMO ontology. A new ontology or extension is not similar.
Figure 3 has a bad layout.
Figure 3b (same for Figure 4b) does not allow readers to understand the structure of the file. It will be necessary to provide a minimal example file as an annexe.
Questions about the Point Cloud 3D object description format mentioned line 254:
Why use this description format rather than specialized ontologies to describe this information?
What information can be stored in this format? Does this format only deal with geometric features or is it possible to add information such as material?
Line 325: Why self-learning ontology approaches are not used here to enhances knowledge ?.
The article needs careful proofreading:
Line 186: " PDC " instead of " PCD "
Line 218: " compnent "
Author Response
Dear Reviewer,
We thank the reviewer for the comments. Please see theattachmentt for our responses.
Best Reguard,
Parkpoom C.
Author Response File: Author Response.pdf
