Mycroft Community Forum

Padatious conversational bot for Winston (yes, we're back!)

First, please redirect me if this isn’t the proper forum. I see that we have a forum for the Adapt Intent Parser, but none for the Padatious.

What I’m interested in is creating a conversational chatbot/skill based on the Padatious engine. Right now, we have several solutions that are Fallback skills that hand off the parsed query to another chatbot process, such as an AIML or chatscript engine. I want to build the conversational bot right into a skill based on the Padatious engine.

Here’s my use case:
I have a small animatronic in which I have embedded a Pi 3B+ with Picroft. The animatronic (Winston, aka upcycled Teddy Ruxpin) runs off a battery pack and is completely mobile for up to 5 hours before needing to be recharged. I want Winston to be able to make small talk like a well-developed Chatscript/AIML bot. Is this possible to do using Padatious? I’m not sure that the logic is sophisticated enough. I also want to incorporate language-based machine learnng. As Winston hears things and is asked questions, he stores this information and becomes more “intelligent.” Being mobile allows me to take Winston on tour.

Is anyone working on this type of skill or project? Can anyone point in the proper direction?

3 Likes

Hey Charles, I was looking for Winston and just saw this.

A few things you might be interested in:

We also have a currently paused project called Persona. It’s a Q and A machine learning algorithm that gives Mycroft personality. Definitely a project we want to get back to, and a favourite of our founder Josh, but there are more foundational projects we need to get ticked off first.

Thanks. I’ve been playing with the AIML fallback skill. I installed a clean AIML set from the ANNA chatbot. The ANNA fileset only had a few errors and missing variables. The AIML files associated with the Alice bot needs to be cleaned up. They throw a lot of errors and are missing many of the variables.

The RASA solution is similar to a Chatscript solution that I’ve worked with. I’m really not concerned about “protecting” my code and running a remote chatbot. I can run a Chatscript engine locally on the Pi and simply point to it. I might be able crib some of the RASA skill code, though, and use it to evolve the AIML fallback skill into an AIML chatbot skill.

If I can get the AIML bot evolved into a true chatbot, I might try to roll it forward to AIML 2.0.