. Project number and Title
 Proposal 1
 Learning to Identify the Controllability of Containers and their Contents using an Extension of
Self-Detection
2. Should this project be considered for the Best Project award?
Yes
3. Should this project be considered for the top 3 project awards?
Yes
4. On a scale of 1 to 10, how would you rate the overall organization/clarity of the project?
8
5. On a scale of 1 to 10, how would you rate the overall project idea?
8
6. On a scale of 1 to 10, how would you rate the overall research contribution of the project
idea, methodology and/or results?
9
Overall, is the project report clear, concise, and well-organized?
Overall, you have done a great job of framing your project in your report. The sections are
organized well and the deliver the necessary content to get a basic understanding of what you did and
how you accomplished it. The introduction thoroughly introduces the territory your research, how your
research wished to add to it, and I felt that I understood what it is was you would set out to accomplish.
The introduction flowed naturally into the body of the report and I had a sense that you were touching
new territories of robotics since, as you reveal, little has been addressed on how self-detection can aid
in the controllability of containers. Controllability seems somewhat of a loose term due to its “-ability”
and it would have been nice if you had defined it for your research. It would have been nice to know
how the other related robotic platforms accomplished their similar container control computationally in
comparison to yours. I like that you clarified why the robot should have learning grounded in its
behaviors—it reveals the importance of test and verification for learning if we are to have intelligent
machines. It would have been nice to see a picture of your color/motion tracking algorithm in use since
it played a large role and ended up being the primary weakness of your findings (you had included it in
your proposal). I was not quite which audience your report was aimed for, but for myself, some topics
that were introduced and utilized that I’m not familiar with, like mutual information, seemed vague, but
I appreciated that you described how it was computed and utilized. I also really enjoyed your
controllability graphs. It really made sense of how your algorithm worked and how the robot was able to
use the extension of self-detection to determine the controllability of containers as demonstrated. In
the end, you end up demonstrating a unique attribute in a robotic system—the robot learns to identify
what objects the robot can control. Since color tracking was seen as a weakness, perhaps you could have
touched on the topic of perception and that it is obviously not through color alone that we learn about
the environment. Great job Shane!
Page 3

Page | 2
How does the project idea and methodology fit within the framework of Developmental
Robotics?
Your project idea perfectly within the framework “container” of Developmental Robotics.
Robot? Check. Learning? Check. Using your algorithms, the robot learns to detect itself. By detecting
itself it is able to determine objects are other and capable of being controlled and by extending the
same algorithm, it is capable of learning what objects that original object can contain/control. It is a type
of tool use. Your methodology is right in line and familiar with all the reading I have been through for
Alex’s Developmental Robotics course. You make certain to state that your robot is a developmental
robot and in being so, is capable of learning.
Describe what you like BEST about the project.
Your paper is very well organized. It reminds me of the many IEEE formatted papers we have
read in Alex’s developmental robotics course. I was never lost and your research is clearly explained. It is
not easy to come across so clearly and concisely in so many pages. In the end, you deliver a novel
developmental robotics finding which seems readily available to be applied and demonstrated in robotic
systems. Taking notes while reading your report, I found myself writing “cool” in the margin either
because you had offered up a new finding or answered a question that was being formed in my mind
while reading.
Describe what you like LEAST about the project.
Besides a few minor typos I found, there was little I found in error. I suppose the only lingering
question I have is how your research is applicable to our future. In other words, why is your research
important and relevant? It seems that roboticists generally expect their audience to accept that it is
good for robots to be self-aware. As a result, it seems it may be necessary to indicate why such a
“power” is necessary for humankind. What does your research offer for the future of robotics? I
suppose it’s a big open ended question, but as someone who is not intimately involved with robotics, I
wonder what the end objective is. Does your research aid in assembly line robots? Does it aid in home
assistance? It is never clearly stated and it is left up to the reader to decide why your project is
important or not.
Do the methods, results and contributions of the final project correspond to what was
presented in the initial project proposal?
There were several items from your proposal that didn’t show up in the project report. You had
spent a bit of time describing deep learning algorithms with neural networks but it is not mentioned if or
how it may have been applied in your project. You do not disappoint however, your evaluation method
still holds the same regardless of the implementation of self-detection and container control algorithms
you utilized. Your findings meet your expectation in that the robot “successfully gains a representation
that unites the controllability of a container’s contents…” I like your term “inside spatial relationship”
but you do not appear to mention this term in your final report. Perhaps the robot does not actually
gain this understanding.
Page 4

Page | 3
Are there any major details left out with regards to the methods, algorithms, or
experimental design described in the report?
To me, other than what I have already previously mentioned, it does not seem that any major
details with regards to the methods, algorithms, or experimental design are left out of the report.
Do the experimental results reported in the paper demonstrate success?
I would say that your results demonstrate success based on your expectations in your proposal:
“The robot will be successful if the robot can accurately construct the controllability image.” While you
do not define what a controllability image is in your final report, all findings show that the robot uses the
image captures to demonstrate behavior-grounded container control.
Do you have any suggestions for improvement and future work?
I would like to see how your algorithm works with fluids. Can the robot learn to pour fluids and
understand how they can be controlled in a container? Future research on such a topic seems
interesting.
How close is the final project report to being publishable as a conference or journal paper
(consider the research papers that were part of the course reading)? What would it take to
get there?
The paper seems very close to being ready to be published as a conference or journal paper. I
did catch a few typos that need to be corrected. You have plenty of sources and seem to have good peer
support. In addition, the idea, as you clearly state, is novel from the get go since little has been done in
this realm. As a result, any findings are worth reporting and so your success is surely worth sharing.