How does it work?
• Sensor at each facing measures amount of light. Light readings are different when covered by product, and when loading inventory vs. removing inventory. • Camera at each shelf - takes a photo and stores locally • Connected coolers: Data from controller is sent to Bluemix at configurable intervals • Occasionally connected coolers: Data from controller is loaded onto mobile device for sending at later time to Bluemix • Data sent in via MQTT for identification by Watson Visual Recognition service • Data is fed to Bluemix application for inventory adjustments
What does it measure?
• Was something added or removed from a row? • What was added or removed. • If inventory was added, then increment; if subtracted, decrement count. • Is what was added according to planogram? if no,flag system that an inappropriate item was added and show picture • Is the row empty (regardless of what we believe the count to be)
Does the retailer/bottler/distributor have to do anything different to make your solution work? (if yes, please explain)
• gravity fed or spring loaded trays • There are no connectivity changes required because we were advised that all sellers have mobile devices to download/transmit data
What materials does your solution require?
• Controller (1 per cooler) • Memory for controller (i.e., SD card) • Sensor (1 per facing) • Camera (1 per shelf) • Cloudant Sync • Cloudant • Bluemix • Watson Visual Recognition
Best guess on cost to implement
• $20 - $40 per cooler
Challenges you/your team ran into
• staying under $20 per cooler
What you/your team learned
• Watson is cool
• Functional prototype • Phased approach (biggest win for least cost)
Anything else you want to add:
Analytics to proactively predict future inventory.
Capable of providing real-time notifications (i.e., when product is removed, stocked, running low, etc)
Social interaction (twitter/instagram/etc) and sentiment analysis enabled