Ricardo Delfín
Ricardo Delfín

Twitcher: The Robot Tweeter

CS378 Class Photo for spring 2017, the class I was a Peer Mentor/Undergraduate TA for.

How does it work?

This project makes use of the BWI Project codebase. Specifically, the bwi_kr_execution, or knowledge reasoner, package. This provides a set of services to send goals to the robots.

The twitter handler is divided into 4 distinct components: The twitter connection, the manager, the interpreter and the action executor. The twitter connection essentially is in charge of connecting to twitter, both directly fetching new tweets and sending responses.

The second, the manager, coordinates all actions. It receives new tweets from the connection, decides when to send responses, and sends requests to both the interpreter and action executors.

Next up, we have the interpreter. This one is in charge of converting tweets into an action. This self-contained box is where all the research will happen going forward. Currently is uses a simple regex pattern, but the long term goal is to use NLP to identify messages accordingly.

Finally, we have the action executor. This one simply translates action messages into goals to be sent to the knowledge reasoner. From here, we can ask the robot to create a plan to go to a particular room or hallway.

Want to look at the code?

Download the code here: