March 14, 2016 | Written by: Mario Kamburov
Categorized: Events | Internet of Things | Watson
Share this post:
This past weekend I was mentoring at a hackathon in Prague focused on the Internet of Things. During the hackathon, most of the teams were playing with RaspberryPIs, Arduinos, Drones, Spheroes and many other gadgets.
The winner of the hackathon was a team of five people who hadn’t slept for three days (Dominik Franěk, Jiří Pětník, Daniel Chabr, Martin Strouhal and Martin Stranný). The project of the winners used an older version of a robot called RoboMe, which is not that expensive (around $50) and was controlled via the microphone jack through a mobile phone. Even using an older model, the team managed to integrate Watson Text To Speech, Speech To Text and Dialog service together into a working product using IBM Bluemix Cloud and Watson capabilities in just three days!
The overall idea is disruptive and innovative way of welcoming clients or guests in smaller, medium or bigger companies instead of receptionists.
The key points the project solves are:
- High costs per employee
- Complicated company spaces and navigation through business and meeting rooms
- Automating the role of receptionists
- Having discussions with handicapped persons.
Where can WelcomeBot be used?
You can impress your guests in:
- Business premises.
How Watson can boost your business
The project dealt with several use cases. One of them is the welcoming part, where the robot is listening to the client’s voice and translates it into text (command) via the Dialog service. Based on that dialog, it starts a flexible conversation with the real person. Although the receptionist may need to guide the clients to specific places, the role of the robot would make their lives much easier. As shown in the video below, the Dialog service was trained for few specific types of events and rooms to navigate:
WelcomeBot could also direct visitors to a specific person, as shown below:
Another problem the robot could solve using IBM Watson services is contacting company employees from the reception area by sending them an SMS or voice call via the Twilio service.The guest comes to the entrance of the building, starts a conversation with the robot, and then the robot processes the name of the person the guest is searching for and sends him/her an SMS or notification:
Future improvements of the project could involve some of the native sensors of the robot, like guard mode being triggered from movement in front of the robot, shaking the hand of the robot to initiate a discussion, etc.
The solution is using fully IBM Bluemix and Watson services. For the backend, the team chose Node-RED flows. The mobile app is communicating directly with Watson STT and TTS services for speech communication. The rest of the flows are processed in Node-RED, for example the dialog and connection to database.
Below is an example Node-RED flow for dialog communication and listening triggers:
If you have any questions feel free to ask or contact me on @_mario_kam_.