How does it work?
The solution uses a touch sensor and distance sensor to determine if there are cans located on a shelf row. Data is collected locally in the cooler and is pushed to the cloud at timed intervals via WIFI or GSIM. The prototype sends data using GSIM when a certain level of light hits the sensor. Also touch data is sent to the cloud via WIFI. The solution will hopefully be built for new cooler deployments and/or retro fitted into existing cooler can shelving systems. The basic solution does assume stores are following the recommended planogram for stocking inventory to use for determining what product has been purchased. The data will be pushed to the Microsoft Azure cloud with integration into their machine learning and/or bot framework.
What does it measure?
The solution measures the distance between the cans and the cooler rear wall to determine how many cans are left on a shelf row. We will also measure if a can is touching the front of the shelf and the number of times the door opens for can access. We will us this information for predictions using machine learning and possible bot execution for distribution/delivery.
Does the retailer/bottler/distributor have to do anything different to make your solution work? (if yes, please explain)
Yes, our comprehensive solution would involve some level of crowd stock verification and/or dynamic truck distribution based on data analysis. Therefore there could be engagement with trucks, RPC and/or pallets management in order to support the solution. The solution will require periodic checking and maintenance of components involved. Also lighting levels would need to be managed in the cooler to display a brighter level on door opening should the light sensor data push be implemented.
What materials does your solution require?
Linkit One or similar board, Redbear Duo or similar board, Touch sensor, Distance sensor, Light sensor ,OLED Display (optional) ,1Sheeld board (optional), SIM card
Best guest on cost to implement
Most of the items involved in the solution could be fabricated in a reduced capacity than what we used out of box. Therefore I guess estimate $150 per cooler for prototyping and less than half of that for rollout.
Challenges you/your team ran into
Determining both cooler stock and product type available without requiring intervention by store stock associates. Identifying the appropriate triggers for data collection and communication. Incorporating the use of existing coke infrastructure across various countries and deployment scenarios.
What you/your team learned
I gained great insight into the coke infrastructure and the challenges of scale for whatever solution is implemented.
Further investigation into solving the challenge of scale from smaller to bigger markets. We feel there are areas where we could reduce the cost of the solution or solve other business problems within what we have developed. Additional software development with the cloud connectivity for bot and machine learning. Also would like to investigate adding features to packaging in order to determine product type by touch or scan.
Anything else you want to add
My team is already working with clients involved in truck load and pallet management who are designing innovative IoT solutions. We think this solution could be part of a comprehensive distribution management system and possible crowd stock management ideas we have. Also we have helped clients build successful IoT Kickstarter campaigns so we do have experience building complete solutions.