Tejas was hugely inspired by a TEDtalk by Byron Gillespie, Engineering Director at Intel, describe how inventory management is the biggest problem faced by the retail industry - but perhaps IoT could solve the problem. With a separate internet enabled device watching all shelves, managers would be able to keep track of inventory and buyer habits.
Christine heard about AmazonGO, a brick-and-mortar storefront for the online retail giant Amazon inc. How would an unmanned store be able to keep order in a shop open to everyone? We decided to tackle this problem to help stores keep track of their shelves and restock them instantly instead of an actual person finding about it at the end of the day.
Mark was excited about the new self-checkout counters at the local Target. It was cool how the weighted plates helped the store keep track of checkout items with precise measurements of each unit's weight. What if this could be applied not just to self-checkout, but facilitating a much more free shopping experience overall.
We took all these ideas and came up with GoShop. A hopeful predecessor for future markets, GoShop takes the human error out of retail and allows customers to shop with speed and convenience.
What it does
Stock Overflow manages retail transactions through force sensors located under each item in the store's inventory.
How we built it
Using Arduino Unos connected to FSR force sensors, we captured the weight of the items on each shelf and the number of items. The Arduino tracks in real time the number of items removed or replaced on the shelf, based on these weight measurements. A Python script listens for the Arduino's serial output and sends http requests to our API on AWS Elastic Beanstalk, which then updates our RDS MySQL inventory database. In addition, the Capital One API is queried when the transaction takes place to either charge the customer in the event of a purchase, or charge the store in the event of a restock. The real-time inventory of the store can be viewed via a web page. A sign in page verifies a shopper's identity prior to entering the store in order to maintain a secure space and keep track of all customers.
Challenges we ran into
We originally intended for the Arduinos to be standalone devices, sending requests using WiFi shields, which ended up being very difficult to authenticate with the university network. Another early plan was to utilize face detection to more accurately determine which customers were completing transactions.
The FSR sensors we used to measure the weights of the inventory items also gave us trouble, as they were not designed to be used as scales. Additionally, the small size of the sensors meant that we had to create a surface on which to place items. We initially attempted to 3D-print these surfaces, but experienced hiccups with our CAD software, and ended up laser cutting them out of wood.
What's next for Stock Overflow
We hope that the Internet of Things can streamline the tasks of business owners, while making the retail experience more secure and convenient for both the consumer and the merchant.