Proposal #16 Best Proposal prize? no Overall organization/clarity of the proposal? 8 Overall project idea? 8 * Overall, is the proposal clear, concise, and well-organized? Yes this proposal is clear and organized. They review their approach, present the direction they'd like to take, and detailed the evaluation criteria for determining success. They present the material in a well defined proposal format including tile page, contents and references. I really liked all of the images and diagrams. It really helps with the flow and clarification of what's being stated. * Does the proposal meet the posted proposal guidelines? It does. It gives history on the topic, clearly states the goal and puts together a framework to achieve that goal. * How does the project idea fit within the framework of Developmental Robotics? It looks at associating multiple sources of data to assist a robot (after training's complete) with navigation within a building/house. Specifically it proposes the concept of using near field communication to allow the robot to reground to it's location when passing within proximity of these devices (RFID). This in addition to current technologies for mapping out a room and movement would be combined into a common data library for later use. The learning part would be a partial challenge for me to say that it's developmental, because it's commanding the robot to do something and then record data. Instead of just letting the robot roam the building/house and learn the locations of all these wireless “checkpoints”. Also the learning would use a predetermined map and locations on the map for the robot to relate to. * Describe what you like BEST about the project idea. I like the fact it's taking a current research project and extending on it using a common platform that others would be able to duplicate and build on. So the Android phone based bots could pickup this work and advance it. * Describe what you like LEAST about the project idea. It's a difficult proposition to talk about a proposed concept and how you'd do it and then reduce the actual implementation and evaluation to a small subset of what was discussed earlier in the proposal. (citing hardware and software limitations...) The project then becomes more hypothetical. * Do you have any concerns about the project? No big ones, it looks like a good project for the time a lotted. It has defined evaluation criteria to determine the results. I'm a little concerned with the small amount of actual coding and all of the manual steps in the evaluation. It sort of reduces the actual project to just a phone taking pictures of barcodes and storing them in a database. Since there is no discussion on using the data in a neural network for making decisions after the learning phase. * Does it seem doable in the remaining time? Yes. * Does it seem too difficult? It doesn't seem too difficult, the project is simulated robot navigation, database lookups, and taking pictures of barcodes. * Are there any major details left out? A discussion on how the data would tie together and be used beyond just a database lookup when RFID tags are scanned. Maybe when the robot is in learning mode it records location information when it scans the tag and also other sensor information about the object. So if it comes into contact at a later point during operational mode, with something that meets the sensor description. It can use a neural network to figure out that this new object is like a previously learned object and it knows where to go to get to that previous object. (Or maybe what to do with it....) The discussion about how they're going to generate or borrow the data sets used in analysis. Weather they're going to generate this data, connect to sensors and gather the data, or borrow it from a previous project that did a similar study. I believe it's also missing the discussion about the type f data set they're going to use to understand how the robot decisions would occur. The reason I believe this is important is because when using a mixed set of data (images, sound, other sensory input, etc) the complexity goes exponentially up and determining success during evaluation could be come difficult. For example if modeling a Apple, you'd have to come up with relationships where the word apple is linked to pictures of apples and other relevant data. This same rough set of data would be needed for all items the robot learns, so when the robot goes operational and tries to determine what a object is, they have the necessary previously formed relationships already formed during the learning exercise. On the other hand if only one sensory input is selected to begin with and say imagery is that input. The learning process can characterize similar data from that one source and more data could be learned to make the evaluation process simpler for this size of project. * Does the idea rely upon technologies that are not currently available? It relies on a couple technologies that aren't available. The proposal talks about how a specific mapping/location application would be a good fit but it's not available. No real other options where discussed for replacing that application, aside from creating a custom application that did a subset of capabilities. It also discusses issues with not having hardware that can do near field sensor reading, but there is a substitute discussed using barcodes. * Do you have any suggestions for improvement? (I caveat this is based on my understanding from your proposal) I'd highly suggest doing something more with the data you're gathering and have part of the evaluation be the decision making process. So that your evaluation is more then just looking up values based on reading a static barcode and reporting them. You could have a robot read a barcode and use the object property data to find other objects that are similar and then have the robot do something with that information. ie. A robot scans barcodes for a bunch of cups laying around a room. If you instructed it to pickup all the cups it should be able to find all the cups based on object information and then navigate to the cups, scan them. * Do you have any suggestions for related work that should be cited? I don't. * Any other comments or suggestions? Good concept. This is a really cool idea to allow a robot to get another set of object data beyond what it's sensors could provide when it encounters a new object.