Today‘s banks are not attractive, due to the limited availability of banks advisors at peak times; the language and cultural barriers between bank advisors and customers; and last but not least, the fact that all people like immediate attention, and prefer personalised services.
What it does
Pepper is a robot with movement, speech and vision capabilities. Empowered by the cloud and a rich dataset, it reveals itself capable of providing a deep, customized, client experience. As such, our project revolves around enabling such an experience, with a prompt reaction to every customer waiting in a bank, improving customer satisfaction.
How we built it
The hardware is a Pepper robot by SoftBank Robotics. Using the built-in camera we recognise a person and identify them against our customer database in the cloud.
- NAOqi API and Choregraphe to control the robot,
- Microsoft Face API to validate your identity using a photo of your face,
- CreditSuisse HackZurich API to store our customer data in the cloud
Challenges we ran into
- Choregraphe uses a visual programming approach which was unfamiliar to us
- Choregraphe seems to be rather unstable and many things cannot be tested in the simulator
- Sloooooooooow WiFi makes cloud request a waiting game, and heavily limits the accuracy of face recognition, since higher resolution images are prohibitively large to be sent over this WiFi
- As there were less robots available than teams working on it, it was tricky to convince Pepper to not run away
- Pepper's camera is not very sensitive in light low environments resulting in somewhat blurry images
Accomplishments that we're proud of
Chaining together several different platforms, envisioned with different goals, and create a flow of data and algorithms that effectively solve a problem in a scalable manner.
What we learned
What's next for Pepped Up Bank
- Implement better, more personalized, customer experience, starting with careful treatment of private data.
- Take over the world. Robots are people 2, you know!