M14 First Skeleton Companions : April 2009

D9.2. First skeleton companions M14 Result of completion of T9.1.2 Report, live experiment

Trying to understand the role of foam in lirec, in order to start working on what deliverables can be for this milestone.

Dissemination:

  • Do we need to start thinking about Lirec in art galleries, art installations - for wider dissemination to the public?
  • How to convey long term companionship in such a format?
  • What about downloadable companions/games for the masses?
  • Downloadable iphone/android/equiv apps?
  • Get these available early - and with source - to start collecting a momentum of interest

Pleo and existing communities of robot hackers:

  • Can we use the pleo as a tool for dissemination to a community with an existing robot framework?
  • Are there any other similar communities? (Lego Mindstorms?)
  • Is the pleo being used in any of the scenarios?
  • Can we get/do we need an PDK for any of this to work?

ION

  • What nature is the development of companions using ION?
    • Java/other code?
    • XML descriptions of personalities etc?
    • Where will this development material live?
    • Will it be shared between partners?
  • Can we use the Pleo with ION?

Some Directions

These are some vague ideas and directions we could head in. I'm not sure whether foam's work is best done as theming and adapting the existing lirec scenarios, or how much we need to/can head in our own direction. I guess the nature of this will become clearer after forthcoming meetings.

I'm assuming foam needs to look at widening the interest of the work done to as big an audience as possible. This means the other research partners can concentrate on the scientific nature of the scenarios/experiments.

We need to consider what access to hardware we will have in the short term, and how far we could get with development for this deliverable. It would be good to have some foam specific goals which can give direction to the development… so with that in mind:

What:

  • Multiple interacting agents
  • Long term memory and personality based on fictional characters
  • Short term memory based on beginning of play
  • Audience participation

Why:

  • Audience familiar with characters
  • Could form a good benchmark for social interaction
  • How do they resolve interesting social situations
  • Start in the middle of a familiar story?
  • What happens after the story ends?

Further:

What:

  • Look into alternative methods of communicating affective state to the user
  • Non-vocal sounds
  • Non-literal visuals

Why:

  • Uncanny valley
  • Remove the problem of simulating human expressions
  • The challenge of meeting the robot halfway in mutual understanding might be more engaging
  • Task of deciphering results in more implied/imagined content
  • People work together to try and understand

Further:

What:

  • Interface to allow the user to easily build and configure robots
  • At the software level
  • Reconfigurable layer 2 ↔ layer 3
  • Need some common goal, or competition

Why:

  • Suspicion that the attachment gained by having a role in the design of a robot is greater than a finely tuned system (or even a system which works)
  • Present the internals of the system to the user in some clear manner
  • Long term attachment through tweakability

Further:

What:

  • Physical dependency
    • Needs feeding
    • Needs transport - has wheels, but asks to be carried everywhere
    • Needs light - not solar powered, but doesn't like the dark
    • Gets cold - needs physical contact
  • Emotional
    • Paranoia - requires constant reassuring noises
    • Anger - needs calming, physical contact
    • Sadness - needs singing to
    • Gets over excited - needs calming down
  • All of the above?!
  • It may be important that they are entirely unnecessary needs

Why:

  • Sounds really irritating, but - Tamagotchi
  • Long term attachment through manipulation of the caring instinct and guilt
  • The feeling that something is dependant on you
  • More effective/complex strategies can be learnt over time - “give him to me, I know how to calm him down”

Further:

Scenarios

How can some of this relate to the existing scenarios? [TODO]

Scenario Hardware Quantity and target audience Notes
HW Team buddy Mobile Robot Multiple users, students Collective memory for a small group of workers, knows who is present, and where people are who are not. Knows about collective events, deadlines and such. Learns personalisation for each team member. Remembers bits of personal information volunteered by group members.
HW Personal guide PDA 1 user, students Portable companion which knows it's location and can act as a guide, and tell anecdotes and stories about places. Includes route planning, knowledge of weather conditions.
HW In the wild Graphical display Multiple users, students A gossip robot, converses with individuals and remembers who they are. Uses a text based chat interface, also simple games included. One of the main motivations is to deal with deliberate rudeness gracefully.
INESC-ID Game companion Static robot or graphical agent 1 user, children and teenagers A companion for playing games which will learn the individuals it interacts with, and their histories of playing together
INESC-ID Personal trainer PDA/graphical display 1 user, not specified A companion which will create and update a fitness program for the user, migrating to different embodiments as required
INESC-ID Welcome to the jungle Mobile robot/graphical display 1 or 2 users, not specified A robot augments a game world which it can migrate into when required, communicating with the user at all times
UH Fetch and carry Mobile robot 1 user, elderly or disabled in the home environment A robot to assist in physical tasks
UH Cognitive prosthetic Mobile robot/PDA 1 user, not specified Organiser and helper with task completion. Offers advice on where objects are, and can check to see if tasks have been completed (objects in correct positions)
UH Telepresence card player Robot/PDA 2 users, not specified Robot as intermediary for a card game, watches one player's cards then migrates or moves to second player and offers advice and encouragement
UH Teaching proxemic preferences Mobile robot Multiple users, not specified Using gesture recognition, the robot learns what distance to be from the individual user.
UH Travelling companion Multiple embodied robots/PDA 1 user, not specified A companion which is able to accompany the user for long periods of time in different settings, at home, while shopping, at work and while travelling.