Project Presentation Immediate Alert Cystem (Team 9) NarrAUTOr Proposal
NarrAUTOr is a reading robot assistant able to help in a variety of situations, whether it is used by a child, a student, or a person with a disability. It incorporates mechatronics, hardware, image processing, and computer vision in order to turn the pages of and read aloud the book you would like to hear.
The setup of the device itself will include multiple components controlled by a portable computing unit; two page scrunching mechanisms that uses friction to create a loop at the base of the page, where the page connects to the spine of the book, corresponding to the direction (forward or backward) the user would like to flip; a page flipping mechanism that flips the scrunched page; a camera sitting atop a stand that points down at the book and snaps a picture of it; and a computer vision program that takes the picture of the open pages, translates the image to text, and reads out the text.
While there are existing devices with similar function to the one we are proposing, our implementation takes the best features of the ones we have seen and combines them into one. The feature that sets our project apart is how we will include an algorithm that determines when after scanning a set of pages to turn to the next set of pages, so that there is no lag in the reading of the book to the user. An additional good feature is that we will be able to flip pages backwards, so that users can review what was previously read; to prevent the need to re-scan pages as the user goes backwards, we will enable Narrautor to save the last few audio files produced from previous pages.
In addition to this post, we have uploaded our initial project report. It is important to note, that this report’s content is actually different from the project description given in the previous paragraph. This is due to our team deciding to pivot from that project idea to this new one last Friday. Our initial project idea that we proposed to the course staff, was a real-time bus detection system that would use computer vision from camera data along Forbes Avenue to find a Pittsburgh Port Authority (PA) bus (specifically any of the 61s) coming to a nearby bus stop and notify users that it was coming within a few minutes. Some explanation of why we pivoted from this idea is noted in the following bullet points:
– learned that there was currently data like this being observed by CMU’s Traffic21 group, so contacted them to see their data, which they told us we could use for this project
– spoke with one of the researchers, saw their data, and went to see the cameras at their location on Friday, September 14th
– The camera data was not at the resolution we needed (was too low) as while we were able to detect a PA bus, we were not able to distinguish what type it was (e.g. a 61A).
– Apparently they were only allowed to be on the open WiFi, but that sometimes lost connection, and they had to consistently make sure that the camera had power.
– While we would have been able to use Traffic21’s data, after seeing it and their cameras, we determined that we would need different cameras, and more of them since we wanted to be able to track buses in advance.
– The cameras they used and the setup of it, was around $200 for one camera (we believe), so if we wanted 3 extra cameras we would have already exceeded our budget in that area with nothing left to spare for other items.
– Overall, while the idea seemed cool, and we were glad to have learned of a group on campus that is doing something similar, we decided that it was not a good idea to implement this semester.
Due to this pivot, we have also uploaded our Project Proposal Presentation, which was given on Wednesday, September 19th to the 18-500 class, and details the idea for project idea NarrAUTOr.