Yakult

Natural language processing

Electronic Engineering

Mechanical Engineering

Overview

Yakult’s Japanese vending machine was another world first – a smart machine that ties in with Yakult’s Japanese heritage and teaches customers Japanaese. For each word correctly pronounced, customers are rewarded with a little bottle of Yakult.

The technology behind the project involved using Natural Language Processing (NLP) and facial recognition to detect customers approaching, engage them, and detect how accurately they pronounced key words, including ‘hello’, ‘good morning’, and ‘thank you’.

Stage 2 of the project involved developing an interactive digital game that customers battled to win, upon which they received a prized Yakult!

Brief

Yakult wanted to create a unique vending machine that taught customers how to speak Japanese, and rewarded them with a little bottle of Yakult when they managed to pronounce it correctly. Adding to the experience, the machine can detect when customers approach it and start talking to them immediately.

  • Online connectivity
  • Facial recognition
  • Natural language processing for Japanese words
  • Temperature monitoring and remote logging

Facial Recognition

In order to make the experience as fluid as possible, customers are automatically detected as they look at and subsequently approach the machine. Once the machine detects the customer’s intent to engage, it automatically starts the experience, and the machine begins to speak to the customer, welcoming them, and inviting them to take part.

  • Embedded camera for facial recognition
  • Embedded microphone for speech detection

Can you make it to the end?

In collaboration with Kino Creative, a second mechanic that proved extremely popular was developed, requiring Yakult aficionados to make it through an ever-faster game to be rewarded with a fresh, healthy little bottle of Yakult.

Interactive Digital Display

The vending machine was a custom build, including a 42” capacitive touch display, enabling a reactive, intuitive interaction between customers and the display.

All interactivity and machine functionality is remotely managed and can be updated as required.

OTHER
RECENT WORK