Project Number 2: Behavior-Grounded Object Identification, Grouping and Ordering by a Humanoid Robot Should this project be considered for the Best Project Award? Yes. Overall, the project idea is novel and heavily researched. The execution is thorough, and used the sample size proposed. The work involved for the proposed project was very ambitious, but the team was able to successfully accomplish it corresponding to the proposal. The final report is presented in an organized and structured format, just like the proposal, closely resembling to the IEEE paper publication standard. It is also well referenced and polished. Should this project be considered for the top 3 project awards? Yes. For reasons please see the answers to the first question. On a scale of 1 to 10, how would you rate the overall organization/clarity of the project? 9.5: The project report, just like the proposal is presented very similar to the IEEE paper publication standard. However, it is missing a few sections (please see Concerns, issues, questions, and proposals section) Reasons: • The Abstract section provides relevant information about the project, defines the project scope approach, and gives a summary of the project results. • The Introduction section cited many well researched references relevant to the project that provided the background information the reader will need to understand the project. It also describes how the rest of the project will be presented. • The Related Work section is divided into two subsections: Psychology and Cognitive Science and Robotics. The first section presented and summarized a comprehensive list of past researches over psychology and cognitive science relevant to the project idea and approach while the second section presented and summarized a comprehensive list of past researches over robotics researches relevant to the project idea and • The Experimental Platform section was presented in a very easy to follow manner. This section described the experiment design in detail and contained the comprehensive knowledge required to set up the experiment. There was very little ambiguity. The figures provided in this section helped the readers to visualize the objects used for the experiment and how the robot detects an object and how data are collected. • The Feature Extraction section is divided into proprioceptive feature extraction, auditory feature extraction, visual feature extraction, hand proprioception feature extraction in each section, how the data and what data were collected was clearly explained. • The Theoretical Model section thoroughly described how the experiments were carried out, how the problems: object recognition, object grouping, object category recognition, and object grouping are carried out and how the data was analyzed. The tables and figures in this section help the reader to quickly see the results and interpret the data. • The Conclusion and Future Work section though rather short, gives a summary of the results from all of the experiments and touched briefly on future works. • The Appendix section gave a brief team member description, presented a comprehensive list of the software packages needed for the experiment and a timeline for the project. • The References section listed all the references used for the project according to a standard publication format similar to the MLA citation style. On a scale of 1 to 10, how would you rate the overall project idea? 9: The project idea is interesting, relevant to Developmental Robotics, and contains novelty. The novelty for this project is definitely publishable but no ground breaking results were found and this idea is based on many other previously similar works for robotic object identification and many projects existed in dealing with robotic object sorting. There were also already existing researches that used behavior-grounded models to autonomously perform object identification, grouping, and ordering. The novelty of the project idea lies within the training stage: the robot will not be told which object it is exploring at any given trial and that the sample size is bigger. On a scale of 1 to 10, how would you rate the overall research contribution of the project idea, methodology and/or results? 10. Aside from the answers to the first question, the amount of work involved for this project is significant and yielded successfully results. The data was carefully analyzed and well explained during the final report. During the project presentation/ demonstration I was able to ask the team about some implementation details and received very satisfactory answers. Concerns, issues, questions, and proposals • Organization/punctuation o There was no VI section; it jumps from V. Theoretical Model to VII. Conclusion and Future Work. The Theoretical Model section should be broken down to a few more sections, such as Experiment Approach and Results. o The Conclusion and Future Works is only a little over a page and there things that could be expanded and explained for this section. For example, the team could have a list of items for future work, and have a paragraph to explain each item. This section could also expand on lessons learned from this project. o Issues should be elaborated and put into a new section. o It was proposed during the proposal that an index terms to be included for this paper is intended for paper publication, but no index terms on the final paper o It would be nice to separate Motivation section from the Introduction section and was brought up during the proposal review but was also ignored o There should be a page number references and was brought up during the proposal but was also ignored o Is there any reason why the references were not numbered in ascending order corresponding to the structure of the project? (For example, when the first reference was cited in the Introduction section instead of using [23] why not start with [1]?) • Experiment details o As brought up during the proposal review, under the section Experimental Platform it described the robot as a humanoid robot with an upper torso and two 7-DOF Barrett WAMs. This makes it sound like the experiment required both 7-DOF Barrett WAMs. However, only one 7-DOF Barrett WAM will be used. This should be clarified. o For figure 11, I am a little confused on how the data is analyzed. Under the description, it states that the similarity matrix W used was for five big stuffed animal and five plastic bottles, however, the data labels on the graph was about metal objects and metal tins links. Also on the graph, the red and blue markers are not distinct; they are rather randomly clustered where someone else might cluster the right two blue markers as the set of red markers. o As brought up before, maybe it is good to incorporate Gibson’s idea of affordances into the project by eliminating characteristics and features not relevant for the humanoid’s way of object classification. This could be a future work topic. o As brought up before, what target audience is this project addressed to? More information could be included on who is interested in the findings of this experiment and potential future audiences.