1. Project number and Title: 13 - Dropping Disks on Pegs: a Robotic Learning Approach 2. Should this project be considered for the Best Project award? (yes/no): No 3. Should this project be considered for the top 3 project awards? (yes/no): No 4. On a scale of 1 to 10, how would you rate the overall organization/clarity of the project report? (1-10): 6 5. On a scale of 1 to 10, how would you rate the overall project idea? (1-10): 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? (1-10): 7 Overall, is the project report clear, concise, and well-organized? The final report comes across as a lot better organized than the proposal. The sections flow really well into one another, are very clear, and are definitely concise. The figures have been shrunk down and better integrated into the paper and all of them contain labels which definitely help with the feel of the paper. The final report stills comes across as being fairly rough and there are several items that just stand out and say look at me. The final report does meet the page requirements as stated in the class syllabus. At only 11 pages it far below the specified 25-30 pages limit. The project group does consist of only one group member so I would expected it to be a little short, but definitely not as much as it is. This matter is only exacerbated by other features of the paper such as the large font size and the paragraphs being double spaced. Best guess is that the final paper has at most 5 pages worth of information. As stated earlier the figures are definitely better than the proposal, however they do still leave a little to be desired. The placement of some of them seems really awkward. For example the figure text of figure 1 just seems out of place either because it is on the side of it just doesn’t line up well, especially with it being so close to the actual figure. Also figures 2, 3, and 4 have spacing issues with figures 2 and 4 missing spacing between the images and figure 3 having no uniform spacing between the figures. The general feeling I get from the final report is that it was thrown together at the last minute, probably as a result of the project implementation not wrapping up until the very last minute. How does the project idea and methodology fit within the framework of Developmental Robotics? The proposal seems to fit within the framework of developmental robotics especially since the main concept is to allow a robot to learn how to place discs on pegs. The dropping of discs on the peg is akin to Alex’s research on learning tools affordance in which the robot uses babbling behavior to determine those actions that are likely to lead to the desired results. Describe what you like BEST about the project? The concept of allowing a robot to learn how to place discs on a peg is great idea. Dropping a disc on a peg is definitely a good starting point for enabling the more advance behavior of actually guiding the disc unto the peg which in turn can be used a jumping point a for a variety of behaviors that require either inserting a object into another (key into a lock) or place an object over another (socket over bolt). Describe what you like LEAST about the project? The project feels under justified and incomplete. It feels like very little effort was expended in trying to build up to why this research is valuable and what doorways it has the potential of opening up. The previous work section is underwhelming to say the least. The research that is presented is pretty much glossed over providing little to no information. Further research that should be referenced such as Alex’s Tool Affordances and Shane’s Controllability of Containers is not. At only 56 trials it feels like the experiment was barely ran. This feeling is further enhanced by the apparent lack of randomness in the trial data. It almost like the project attempted to run the experiment to completion without validating that everything was setup correctly. Do the methods, results and contributions of the final project correspond to what was presented in the initial project proposal? The implemented project and the final report hold wells with what was presented in the project proposal or at the very least there does not appear to be any obvious deviations. Are there any major details left out with regards to the methods, algorithms, or experimental design described in the report? The report seems to do a fairly job of laying out the project and how it was implemented. If anything could be considered missing it would be the experimental design. The final report does a really good job of defining what a trial success is, however it doesn’t feel like this was ever related back to how this will help the robot to place discs on a peg. Basically it does not attempt to validate that the collected information can be shown to improve the robots ability to consistently place a disc on a peg. Do the experimental results reported in the paper demonstrate success? The results demonstrated in paper show that the experimental setup can be used to gather data. The project does not take the last step and attempt to validate that the gathered results can be used to improve the robots capability of placing a disk on a peg. Do you have any suggestions for improvement and future work? There are several areas for improvement and future work: * More trials with either randomly generated drop points or success driven drop points * Validation that successful trials can be used to improve the robots capability to drop a disc on the peg and can still be applied even if the peg is moved to a different location * Enhance the experimental setup to include placing the disc on the peg. Show that the dropping experiment can be used a precursor to the placing experiment in order to allow for a greater success. * Enhance the experiment to include multiple pegs and multiple pegs. Make it such that the robot goal is to move a disc from one peg to another. This could eventually be expanded into the robot learning how to do the Towers of Hanoi. * Expand the experiment to include different size pegs and holes. * Expand the experiment to include different shaped holes and pegs. Depending upon how the experiment is ran it may be necessary for the robot to be able to set down and pick up the disc such that it can reorient it. * Switch the experiment around and have the robot place the peg in the hole. All the above variations could also be applied to that experiment. * Expand the experiment to placing a socket over a bolt. If you want to get really fancy you could start by placing the bolt in the hole. Bonus points if you can get the robot to screw in or unscrew the bolt. The goal would be to show that the above research led to a model that allowed the robot to learn this really advanced and general behavior. 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 project in its current form is not publishable, however it definitely has potential. There several items would need to be addressed in order to get it there. The project introduction should be enhanced to show that this project has the potential of opening many doors for further research. The previous work section should be expanded to provide a little more in depth background and to include the research that was listed above. The number of trials should be expanded and the dropping algorithm should be verified as being random or at least be shown to be driven by successful trials. The outcomes of the trials need to be shown as improving the robots capability of dropping the disc on the peg.