Inspiration
Well after hearing about the NCR challenge. My team started to work on extracting from their database. After doing so, and finding thousands of grocery items. We wanted to make grocery shopping easier for all by having the option of buying grocery items from your home via voice through google mini.
What it does
This basically allows the consumer to speak to their google home mini, and ask it anything as long as it has the keyword grocery item- such as "Can I buy a bottle of coke?" coke would be captured by the Google API because we fed the entire list of NCR catalog items into the Google API: api.ai. Then the mini would respond something like Ordering from the NCR catalog.
How I built it
First, we extracted the data from the NCR Catalog using Python, and managed to place it into an excel file. We then took those thousands of grocery items, and put them all under the 'entity' of products in the google API, so that whenever any word from the catalog is called, the google AI would recognize it, and respond accordingly.
Challenges I ran into
We ran into a lot of challenges. Firstly never having worked with a google home mini before to get it paired itself was a struggle, then after realizing that this device must be connected to the wall at all times to function, and is paired via gmail, made it easier to integrate with the Google API diagonalflow, Having the system take the keyword the user says, and parse through the NCR database was also very tricky. We also wanted to connect to virtual reality and pass that keyword over to the oculus, so that it could be projected; however, due to time and tons of problems with the HDMI cabel connection,we were not able to do so.
Accomplishments that I'm proud of
Well, actually extracting all the data from the NCR catalog successfully, and have it all displayed was a real feat of accomplishment. Then having the Google Home Mini, respond correctly after hours of training, was overwhelmingly satisfying. Overall, although we were not able to incorporate virtual reality like our original plan, just having a functioning programmed google home mini that can parse the database is in itself a great task.
What I learned
I have learned a lot from this entire process. My team was the ones who did the actual data extraction from the NCR catalog using python. I am not familiar in python; however, this enabled me the chance to get more exposure, and learn a lot about python frameworks, and also json incorporation. I learned how to merge entities, and train the Google AI to recognize keywords and respond to the user. Overall, I think this skill set I've achieved from this program can allow me to program almost anything on the Google API using cloud platform, and successfully integrate it with the Google Home Mini, assistant or other devices.
What's next for MINI Grocery Helper
Like I mentioned, I would really like to incorporate virtual reality into the program. I would like the keyword picked up by the Google Home Mini to parse through the NCR database, and send to the Oculus, where it can be appear in visualized form to the consumers. Then the consumers should be able to rotate the grocery item completely, and see nutritional data and analytics beside the item so that they can make an educated decision.
Log in or sign up for Devpost to join the conversation.