Page 254 - Designing Sociable Robots

breazeal-79017 book March 18, 2002 14:24

Grand Challenges of Building Sociable Robots 235

13.3 Grand Challenge Problems

The ultimate challenge for a sociable robot is to interact with humans as another person
would and to be accepted as part of the human community. In chapter 1, I outlined several
key ingredients of building robots that can engage people socially in a human-like way.
This list was derived to support several key attributes of human sociality. These ingredients
address the broader questions of building a socially intelligent artifact that is embodied and
situated in the human social environment, that exhibits autonomy and life-like qualities to
encourage people treat it as a socially aware entity, that perceives and understands human
social behavior, that behaves in a way understandable to people in familiar social terms,
that is self-aware and able to reflect upon its own mental states and those of others, that
learns throughout its lifetime to increase its aptitude, and that continually adapts to new
experiences to establish and maintain relationships with others. Some of the grand challenge
problems are derived from these target areas. Other challenge problems address issues of
evaluation, understanding the impact on the human who interacts with it, and understanding
the impact on human society and culture.

Anima machina As the term anima machina suggests, this grand challenge problem
1
speaks to building a life-like robot. This challenge encompasses both the construction of
a robot that can manage its daily physical existence in human society, as well as the design
of the synthetic nervous system that “breathes the life” into the machine.
With respect to the physical machine, overall robustness and longevity are important
issues. Fortunately, advancements in power source technology, actuator design, sensors,
computation hardware, and materials are under way. Improvements in power source size,
weight, and lifetime are critical for robots that must carry their own batteries, fuel, etc.
The ability for the robot to replenish its energy over time is also important. New actuator
technologies have more muscle-like properties such as compliancy and energy storage
(Pratt & Williamson, 1995; Robinson et al., 1999). Researchers are looking into better
mechanisms that approximate the motion of complex rotational joints, such as shoulders
(Okada et al., 2000), or that replicate the flexible movement of the spinal cord (Mizuuchi
et al., 2000). Improvements in current sensor technologies, such as developing cameras
that lend themselves to a more retina-like distribution of pixels (Kuniyoshi et al., 1995) or
increasing the sensory repertoire to give a robot the sense of smell (Dickinson et al., 1999),
are also under way. New materials are under investigation, such as gel-like actuators that
might find interesting applications for synthetic skin (Otake et al., 1999). Cross-fertilization

1. Rod Brooks takes poetic license to convey this idea with the phrase “living, breathing robots.”