Hypothesis: Robot are active in an unknown environment for the execution of a set of actions according to some goals.
That is a crucial experience, allowing the on-line merging of two adaptive systems. The two Aibo's robots are running and we set up a wifi link between them. We begin an on-line association process between their agent's organizations, at the morphological level, with modification of the two organizations, and leading to three specific cases:
One robot is the master and the other the slave, about the current generation.
The two robot's generate a new morphology expressing an integrated new local brain with a new original focal point.
The aggregation is impossible: the two organizations have nothing in common and remain separate. We can apply this aggregation principle to n systems, which is the case of the large area survey.
A video with two current thoughts, a on-line fusion (agent's flux) between two artificial brains for a group decision-making:
Brain A Brain B
A video with goals monitoring.Brain's goals are synchronized:
Here we present a video representing a mental state deteriorated by the emotional system (a minimal scenario).
Hypothesis: The robot is active in an unknown environment to execute a set of actions according to goals.
In our case, an entity is described as a machine with a body and an artificial brain which is not embedded for performance reasons. This approach follows that of Damasio (Damasio, 1995). Contrary to Descartes (Descartes, 1997) who states that the mind is completely separate from the body, the body and mind process in synergy. It is only with the body that the mind can treat different information in an environment. Emotions are in the ontology and are linked to different stimulus.
Here, emotions are processed at the agent level. But in the future, it will be on another level:
The emotions are periodic movements of aggregations of structuring agents.
These aggregations express at the same time the character of the emotion and those of a representation of something (of an object) which is faded.
The type of deterioration represents the type of the emotion.
Deterioration is a modification of emergence making be distinguished an object (focal point).
The intensity of the emotion is represented by the frequency of aggregations.
It is a first step, we have the model to process complex emotions and feelings in the direction of Damasio.
Goals
To play with the Ball or the Bone.
Emotional Features
Pleasure and Dissatisfaction
Acquaintances
For the Ball: physical capacities, persons, specific space, verbs and colors.
For the Bone: physical capacities, an emotion and a color.
For Pleasure: objects (ball, cushion, cover and bone), a verb (to play), a color (pink) and a person (Alain).
For Dissatisfaction: a verb (to work), an object (generator), a state (fatigue) and a person (Alain).
Results and System Behavior
All results are private and this presentation not exposed all crucial points.
Emotion Pleasure is more processed that the emotion Dissatisfaction because it has more accountancies with the context of the environment.
The robot modifies its behavior when the emotion Pleasure decreases. This modification is radical when the emotion Dissatisfaction increases. The robot thinks to play with lot of things when emotion has more and more important, it is submerged by emotions and forgot its aims.
A video without emotions. The robot want to play ball with Alain and has pleasure, so it succeeds its goal:
A video with emotions. The robot doesn't want to play ball, it has dissatisfaction because there is lot of emotions such as sorrow, joy and love. The robot forgot its aims:
Here we present a video representing an artificial mental card with an emergent activity.
Hypothesis: the robot is active in an unknown environment to execute a set of action according to goals.
We use Aibo recognition for objects, colors and persons. For the video, the robot recognize these entities in the test environment: Alain, Mickael, Ball, Bone, Battery and Pink (two persons, two objects and one color, with a recognition rate of 90 per cent), and respect the four phases for the interpretation of data:
Data transit in the multi-agent system. At this phase, it is impossible for a human to interpret information, the behavior of the system.
System interpretation of information in order to create an emerging form with morphology according to goals.
Choose a specific form with morphology to direct the system.
Create an action plan with the emerging form.
During the processing, we name a focal point a set of emergent knowledge evolving in real-time according to sensors values. It is possible to modify the robot memory in real-time.
To realize the video, we had a specific configuration:
An eleven mega octets ontology (with verbs, persons, objects, simple emotions, mixed emotions, colors, physical capacities etc).
Acquaintances between knowledge representing the robot experience.
More than one thousand agents.
More than seven thousand threads.
All Aibo sensors and effectors (Sensors values evolve in real-time).
A simple goal: to have pleasure (linked with Ball, Play, Alain, Cushion, Pink, Cover and Bone) or to have dissatisfaction (linked with Work, Generator, Alain and Fatigue).
A PowerBook G4 1,33 Ghz with 1,2 gigabytes.
In the video (.mov), there is five windows:
Window about the agents (with seven threads per agents) to show the organization complexity. At this point, it is impossible for a human to interpret information, the behavior of the system.
Window about ontology and metric relations between basic concepts values on matrices to explain the variations.
Window of the emergence of the focus point: an emergent group of agents, a "thought".
Window to expose the constraints: numbers of agents and number of sensors and effectors.
A terminal with logs concerning orders sent to agents to direct the system towards goals.
At the beginning of the video, we have two windows. The window on the left is a terminal which shows messages concerning the control of the multiagent system. The window on the right is the focal point of the system, the "thought" which evolve in the time line according to inputs, knowledge, experience and goals. In this window, a yellow circle is an agent group, ie a set of agents which emerge in a same time. Each agent can have acquaintances with agents in its group or with agent in other groups. These acquaintances are the experience of the robot.
Step at time 17 s: we see the window with the system mental card (named paloma in the video, the project code name). Each point represents an agent. Metric system used to place the points is an algorithm building groups of dots according to the different roles. Currently, it is a static representation, there no animation on this window in the video.
Step at time 32: we see the window named "Entities Configuration" which expose the system parameters, ie, agent number, role number, goals, sensor number, effector number and network configuration for the robot.
Step at time 40: we see the window named "IHM" which expose the employed method to modify in real-time the acquaintances between knowledge, so the robot experience, with a matrices stacking.
Let us now explain the set of figures presented in the focal point window during the processing. There is three important sequences:
Between the second 1 and the second 14, three knowledge emerge on the focal point: "Ball", "Bone" and "Sympathy". Elements "Ball" and "Bone" are in the system inputs, the scene, so it is a direct emergence. But for "Sympathy", it is an experience. In the past, the robot had sympathy for a person linked with a ball or a bone. During this period, the robot has a physical behavior because "Sympathy" is linked with physical capacities.
Between the second 15 and 18, the focal point is directed towards "Bone", we note that this knowledge is important in the robot experience.
Between the second 19 and 59, there is a set of figures which show the importance of the roles "Bone" and "Sympathy". This importance is completed by a new role: "Front of" with another set of figures between the second 50 and 112. During the processing the robot idea is more and more precise, just like the robot physical behavior.
It is the last public presentation according to the system features. There is a set of patents concerning several methods used to create the system. The site will be updated with new experiments related to the features presented in this page. This video was created with Snap Pro X in parallel of the artificial brain processing.
