IEEE Robotics & Automation Magazine - December 2012 - 55

Another aspect open for discussion is identity management. Although numerous algorithms exist to authenticate the
identity of specific robots/computers on the basis of digital signatures and cryptographic networking protocols, robots have a
better chance of fitting in human-dominated environments if
they utilize a humanlike approach to identity management,
which is advocated in this article. Other applications include
detecting cheating in games on the basis of assistance from AI
software, providing visual and behavioral search capabilities
for virtual worlds, and making it possible for scientists in fields
as diverse as biology, communications, and e-business to
securely communicate with intelligent assistants and robots.
Development of methods presented in this article not only
provides a broad base for future solutions in those domains
but also opens up new issues related to differences in human
and robot behavioral profiling, collaborative environments,
features, and unique specifics of virtual environments that
need to be taken into account in real-time methodology integration and testing.
Conclusion
This article introduced a new subfield of security research,
which transforms and expands the domain of biometrics
beyond biological entities to include software and hardware
robots, which are rapidly becoming a part of the modern
society. Artimetrics research builds on and expands such
diverse fields of science as forensics, robotics, stylometry,
computer graphics, and security. This article presented a solid
motivation for security research in the field of robotics and
cyberworlds, including six scenarios for automated matching
algorithms followed by the comprehensive survey of robots
and methods for robot dataset creation. Description of visual,
behavioral, and multimodal artimetrics constituted the core
of methodology, with applications and implications of this
emerging area further outlined.
The presented research into trustworthy authentication
of robots will make society safer and better prepared for
the accelerating integration of intelligent technologies into
everyday life. Potential benefits come from the applications of the developed algorithms in many diverse areas,
for example, recognition of military robots, preventing
malicious intelligent software from obtaining access to
resources, securing communication between different
intelligent agents in virtual communities, determining
authorship rights to the results of computation and creative output produced by an AI entity, and identifying
semiautonomous software tools used by hackers.
Potential directions for future artimetrics research include
the investigation of other visual and behavioral approaches to
robot security on the basis of the appearance of new characteristics and abilities in the robots of tomorrow. Even today, it
would be possible to expand robotic biometrics beyond faces
and vocabulary to intelligent software agents, which mimic
higher-order human intelligence. Some examples are provided
below, but the list will unquestionably grow as our success with
AI technologies progresses and we obtain programs, which are

as creative and unique as human beings. We already have programs capable of composing inspiring music [26], drawing
beautiful paintings [25], and writing poetry [14], and limits to
the known abilities of machines are continuously being
extended. It is already technologically feasible to look at profiling text-to-speech software on the basis of voice recognition,
translation software on the basis of linguistic signatures, and
authentication of game-playing bots on the basis of the strategies employed as demonstrated by the authors [117], [119].
Some other open problems are generation and evaluation
of the quality of virtual entity databases and emerging
research on face recognition in the virtual world. Such issues
as emotion recognition, face recognition in the presence of
aging, various geometrical underlying models, different
approaches to virtual face representation, and displaying
options provide a broad variability of avatar faces. Direct
comparison with human face databases and testing the performance of recognition approaches on such databases versus
human databases are an exciting unexplored domain of
research. Evaluating the degree of variability of avatar databases is another open problem.
It may also be possible in the future to profile different
search engines on the basis of the results they produce. Pattern recognition algorithms such as those used for optical
character recognition or for biometric recognition can be profiled on the basis of the error rates. Artificial life and computer viruses can be tracked on the basis of their behavioral
signatures, and game characters on the basis of a combination
of visual and behavioral traits. As hardware robots continue to
improve in their humanlike abilities and appearances, potential physical biometrics worthy of examination may include
gait, keystroke dynamics, signature, and body part geometry.
Acknowledgment
The authors would like to thank Dr. A.K. Jain for his critical
comments in the preparation of the manuscript and Dr. P.
Corke for his diligence during the manuscript revision phase.
The authors would also like to thank the NSERC for its support in this project.
References
[1] K. Ahmadian and M. Gavrilova, "Transiently chaotic associative network for fingerprint image analysis," Int. J. Neural Netw. World, vol. 3, no. 20, pp. 389-403, 2010.
[2] L. V. Ahn, M. Blum, N. Hopper, and J. Langford, CAPTCHA: Using Hard
AI Problems for Security, Eurocrypt. New York: Springer-Verlag, 2003.
[3] S. Ajina, R. V. Yampolskiy, and N. E. B. Amara, "SVM classification of
avatar facial recognition," in Proc. 8th Int. Symp. Neural Networks, Guilin,
China, May/June 2011, pp. 132-142.
[4] N. Ali, M. Hindi, and R. V. Yampolskiy, "Evaluation of authorship attribution software on a chat bot corpus," in Proc. 23rd Int. Symp. Information, Communication Automation Technologies, Sarajevo, Bosnia, Oct. 2011, pp. 1-6.
[5] H. Asoh, S. Hayamizu, I. Hara, Y. Motomura, S. Akaho, and T. Matsui,
"Socially embedded learning of the office-conversant mobile robot jijo-2," in
Proc. 15th Int. Joint Conf. Artificial Intelligence, 1997, pp. 880-885.
[6] H. S. Baird and J. L. Bentley, "Implicit CAPTCHAs," in Proc. SPIE/IS&T
Conf. Document Recognition Retrieval XII, San Jose, CA, Jan. 2005, pp. 191-196.

december 2012

IEEE ROBOTICS & AUTOMATION MAGAZINE

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - December 2012