1) Proposal 1
2) no (Stoytchev’s student continuing own work)
3) 10
4) 10

I think this paper is great, there aren’t many criticisms I can make. This proposed work shows a great

deal of technical understanding and foresight and it is clear that you’ve spent a lot of time developing the

overall structure of the paper. I’ll go through what I liked and what I think you could work on in your

proposal, I don’t think there are any major changes needed.

The things I liked best about your paper are your overall approach and your experimental setup. I

think the study of containers is a very important area of research in developmental robotics, and I think

your approach of determining what is and what isn’t a container based on self-detection is inspired. I

think the three behaviors that the robot will perform are just enough to give sufficient data, while still

being manageable. I also like collection of objects is versatile, though I might avoid that one that looks

like a container, as it might give confusing results having a container in a container. Furthermore, I feel

that with many of the proposals, perhaps mine include, it takes a bit of a stretch to see the connection with

developmental psychology. With yours, however, the connection is very clear and you’ve made a great

connection between what has been done with children and what you plan to do with a robot.

Where the paper fell down for me is where you speak about your methodology, especially when

you go into the algorithms you plan to use. It’s clear that you know what you’re talking about, but it’s

not completely clear to me what you’ll be doing. Consider including a better high level description of

what the algorithm is doing, so that a layman can understanding how your methodology fits together,

even if he doesn’t understand the math behind it. Also, what was the thought process behind having the

containers and the non-containers be the same objects. I understand that this both simplifies the

experimental set-up and the features you detect for determining containers will be object independent, but

there are many non-containers that can’t be containers in any orientation that you may want to include in

future work. Furthermore, this probably won’t be a problem with you, but the work does seem overly

ambitious for the course. You didn’t go into detail about how much work you’ve done already, or how

long you’ve been preparing your research. For the time remaining in this course, your proposed work

seems a little daunting, though if you’ve prepared before, I’m sure you can do it.

Not to overstate my approval of your paper, but I think this proposal is exactly what the

assignment asked for. It’s clear though that this is an adaptation of the work you’re putting together for a

Masters or PhD, so you really can’t be judged against other projects in the course. With that in mind, I

can’t recommend you for the best paper award, though I’m sure you’ll be considered for the much more

prestigious grants that your after.