Artificial Affective Agent - Python Open Source - Raspberry Pi
Source Code will be published 1st of April 2016
The Advanced Affective Agent program links together the EPU emotion processing unit, AI and various IO devices to form an intelligent agent, AI or toy with emotions.
robot = Robot( input_method = 'voice' )
robot.setLights( Bridge('<Philips bridge ip>') )
while True:
response = robot.listen()
if response == 'bye':
breakHundreds of millions of people communicate daily their desires, their thoughts and their emotions to intelligent machines like Google, Facebook or Siri. We predict that before the end of this century humans will talk more to machines than to other humans. Emotion synthesis and the use of emotions remain a fundamental need for humans that cannot be addressed by today’s AI technology.
Imagine intelligent machines and Text to Speech engines able to modulate their voice by feeling what they say, robots with facial expressions and body language directly controlled by the level of their feelings instead of scripts, smart toys, cars and robots developing their own positive emotional personality learning from humans interactions. We invite you to take part in the future; in making the world a better place by giving intelligent machines empathy for humans to bring about a positive future.
The growing presence of robotics and AI in society at large means that meaningful emotional interaction is core to removing the barrier to widespread adoption. Patrick Levy-Rosenthal
This project will allow anyone to create his own AI, toy or even robot with emotional awarness in less than 15 minutes.
The following modules are used in script that need to be installed:
import requests
from lxml import html
import pyttsx
import speech_recognition as sr
import wikipedia
from googleapiclient.discovery import build
import httplib2
from phue import Bridge
from PIL import Image, ImageDrawEmo - class to work with EPU device
Emo.__init__(host, port) - perform connection to EPU devise
Emo.getEmo() - get current emotions of EPU. Returns a dictionary with emotion channels and value for each channel. Dictionary keys:
'channelEXCITED'
'channelCONFIDENT'
'channelHAPPY'
'channelDESIRE'
'channelTRUST'
'channelFEAR'
'channelSURPRISE'
'channelINATTENTION'
'channelSAD'
'channelREGRET'
'channelDISGUST'
'channelANGER'
Emo.getEmoStr() - get current emotion and interpretes it into text string. For example 'fine', 'embarrassed', etc
Robot - main class that links together emotions, AI, input, output and other function as an intelligent Toy.
Robot.__init__(input_method) - initialize the intelligent agent object.
Input_method can be 'voice' or 'type'. When running on Raspbery Pi should be always 'voice'.
Also TTS and Emo object are initialized here.
Also robot and human names are asked and remembered here.
Robot.say(message) - say messages via TTS and also prints to stdout.
Robot.sendAI(message) - send message to AI bot, send answer to EPU (Emo) and reply with the answer.
Robot.listen([param]) - listen what user types or says and starts some scenario depending on what was said.
If input message doesn't imply any scenario, it's forwarder to AI.
Robot.setInputMethod(input_method) - set robot input method. Input_method can be 'voice' or 'type'
Robot.inputMethod() - return robot input method.
Robot.setEmo(emo) - set robot Emo object.
Robot.emo() - return robot Emo object.
Robot.setTts(tts) - set robot TTS object. Now it can be only pytts.init()
Robot.tts() - return TTS object.
Robot.setLights(lights) - set lights. lights parameter for now is only phue.Bridge object.
Robot.robotName() - return robot name.
Robot.humanName() - return human name.
Robot.getLastInputMsg() - return last message sent to robot.
Let people know how they can dive into the project, include important links to things like issue trackers, irc, twitter accounts if applicable.
Based on Apache license: https://github.com/emoshape/TheSmartToy/blob/master/LICENCE.md