|
|
|
|
A human understands people or objects through interaction. The
more and longer they interact, the deeper the human understands
the other. Long interaction can result in attachment and desire
for further interactions. It may also result in boredom.
Interaction stimulates humans, and generates motivations for
behaviors. There can be cases in which behaviors are not
rational.
Objects with which humans interact include natural objects,
animals and artifacts. Studies on interaction between human
beings and animals show positive effects on psychology,
development of children, and so on. Artifacts that affect people
in mentally can be called "aesthetic objects". Such
effects are subjective and could not be measured simply in terms
of objective measures such as accuracy, energy and time.
Machines are also artifacts. Different from the aesthetic
objects, machines have been designed and developed as tools for
human beings while being evaluated in terms of objective
measures. Machines are passive basically because human beings
give them goals. Machines will not be active as long as they are
tools for human beings.
However, if a machine were able to generate its motivation and
behave voluntarily, it would have much influence to an
interacting human. At the same time, the machine should not be a
simple tool for humans nor be evaluated only in terms of
objective measures. Subjective evaluation is important. For a
human, multi-modal stimulation should be influential. People
interacting with the machine or observing the interaction may
consider the machine as an artificial creature. Behaviors of the
machine may be interpreted as emotional.
There are many studies on human-machine interaction. Here, we
don't discuss studies on human factors in controlling machines
used as tools. In other studies, machines recognize human
gestures or emotions by sensory information, and then act or
provide some information to the human. However, modeling gestures
or emotions is very difficult because these depend on the
situation, context and cultural background of each person.
Concerning action by a machine toward a human, an artificial
creature in cyber space can give only visual and auditory
information to a human. A machine with a physical body is more
influential on human mind than a virtual creature.
Considerable research on autonomous robots has been carried out.
Their purposes are various such as navigation, exploration and
delivery in structured or unstructured environments while the
robots adapt to the environments. Also, some robots have been
developed to show some emotional expressions by face or gestures.
However, even though such robots have physical bodies, most of
them are not intended to interact physically with a human.
We have been building pet robots as examples of artificial
emotional creatures since 1995. The pet robots have physical
bodies and behave actively while generating motivations by
themselves. They interact with human beings physically. When we
engage physically with a pet robot, it stimulates our affection.
Then we have positive emotions such as happiness and love or
negative emotions such as anger, sadness and fear. Through
physical interaction, we develop attachment to the pet robot
while evaluating it as intelligent or stupid by our subjective
measures.
Science and technologies have been developed through
objectivism. Because of this, people can share and use their
scientific and technological knowledge in common. When we design
machines, we need to use such objective knowledge. A machine
which has high value evaluated in terms of objective measures
such as speed, accuracy, and cost is useful as a tool for human
beings, especially for automation.
A machine that interacts with a human is not always evaluated by
such objective measures. People evaluate a machine subjectively.
Even if some machines were useless in terms of objective
evaluation, some people put high subjective value on them. Such
machines could be considered as aesthetic objects.
When we design robots that interact with human beings, we have to
consider how people think of the robots subjectively. This paper
deals with pet robots to investigate subjectivity for designing
robots friendly to human beings.
There is an enormous number of studies on emotions. Also, we
can make models of emotions by observing many people. However, we
can not say which model is correct or even the best. We have many
words to express our own emotions, but we don't have the same
definition of internal states of our bodies. Emotions are evoked
in some situation, and depend on context and cultural background.
Therefore, it is difficult to establish a general model of
emotions. For example, if a subject and another person are
interacting, the person's interpretation of emotions of the
subject is not always the same as that of the subject himself.
Even if they had long relationship, they would interpret the
emotions in different ways.
We are taking a position that emotions emerge through interaction
with the environment as Picard classified research on emotions.
There is some research on emergent emotions. Toda emphasized the
importance of studying whole systems including perception,
action, memory, and learning. He proposed a scenario with a
fungus eater to illustrate how emotions would emerge in a system
with limited resources operating in a complex and unpredictable
environment. Toda's robot has the goal of collecting as much
uranium ore as possible, while regulating its energy supply for
survival. The robot has rudimentary perceptual, planning, and
decision-making abilities. Toda proposed urges that are
motivational subroutines linking cognition to action, and argued
the robot would be emotional with the urges. The urges are
triggered in relevant situations and subsequently influence
cognitive processes, attention, and bodily arousal. An observer
of the robot would interpret that the robot's behaviors are
emotional. Pfeifer implemented urges in a mobile robot and showed
emergence of emotional behaviors through interaction between the
robot and its environment.
Braitenberg explained emergent emotions by means of his simple
mobile robot, which had two light sensors and two motors. When
the robot sees a light source straight ahead, the robot moves
toward it, and bangs into it, hitting it frontally. When the
source is not straight ahead, then the robot turns and moves so
that it still approaches the source and hits it. An observer
could interpret the robot's behavior as aggressive as if the
robot felt a negative emotion toward the light source. When
sensors and motors are wired so as to linger near the source and
not damage the robot, the behavior could be interpreted as a more
favorable emotion. The robot could have different behaviors in
the same environment when its internal system is changed. These
mean that emotions have emergent properties and depend on
interaction with the environment.
Brooks argued that situatedness, embodiment, intelligence and
emergence are key ideas of behavior-based robots. The key idea of
situatedness is that the world is its own best model. The key
idea of embodiment is that the world grounds the regress of
meaning-giving. The key idea of intelligence is that intelligence
is determined by the dynamics of interaction with the world. The
key idea of emergence is that intelligence is in the eyes of the
observer.
When a human and a pet robot interact with each other, they
stimulate and affect each other. We call this 'coupling'. When
the human evaluates the robot, the human is an observer and a
subject at the same time. Following Brooks' ideas, the
intelligence of the robot could be determined by the dynamics of
interaction with the subject and environment. Also, the subject
would interpret or measure intelligence of the robot with his own
eyes. The subject's interpretation depends on his knowledge and
experiences related to the robot and its designer. Therefore, the
robot's intelligence depends on the subject's intelligence.
At this point, we don't have an explicit definition of
intelligence. However, as Minsky suggested, we doubt whether
machines can be intelligent without emotions (which doesn't mean
emotion models but emotional appearance of behavior in observer's
view) Therefore, we consider that emergent emotions are key for
intelligence.
As we design pet robots as artificial emotional creatures, an
interacting human does not give them goals nor tasks. Pet robots
are allowed to generate their own goals and motivations for
survival in the world. Therefore,
contrary to Asimov's "The Three Laws of Robotics",
These allowances are the key to let people interpret that pet robots are like living creatures.
A few kinds of sensors and two actuators for mobility are given. Mother board has a simple 6811 CPU. A human operator does not give aims of behaviors. The robot generates its motivations of behaviors by emotions activated from stimuli of its environment. The emotional behaviors are emerged though any explicit emotional functions are not given.
Several kinds of sensors, including vision, auditory and
tactile sensors, were given. The pet robots have learning
abilities for the sophisticated (secondary) emotions through
interaction with human. They appeared in the cyber space and in
the physical space.
We investigated subjective interpretation of robot's behaviors in psychological experiments, in which a picture of a dog was equipped with a 1 DOF tail and subjects were asked to interpret emotions of the dog by wagging tail. Then, a simple tactile sensor was added to the system and the tail wagged depending on stroking the tactile sensor by subjects. In the first experiment, subjects interpreted meaning of wagging by visual and auditory information. In the second one, subjects had tactile information as well as vision and audition. Interpretations of emotions were various because of knowledge of dogs; for example, some had experience of owning dogs. However, the second experiment was much more impressive for most subjects because of physical interaction with tangibility.
 Seal Robot
|
 Cat Robot(with OMRON) |
 Inside |
 Inside |
 Side |
 Sleeping |