Other Artificial Animals

The title of this dissertation emphasizes the generality of our artificial animal approach to computer animation, rather than its specific application in this thesis to fishes. The main components of our approach--modeling form and appearance, biomechanics, locomotion, perception, and behavior--carry over to the realistic modeling of other animals, although the details of each of these components may, to one degree or another, be animal specific. Consider, for example, the design of a realistic artificial cat.

• It is important to capture natural feline form and appearance, but clearly cats and fish differ dramatically, so different 3D models and textures would be needed to model a cat.
• Cat biomechanics obey Newton's laws of motion, as do fish biomechanics. Quite unlike the highly deformable fish body, however, an artificial cat would require the biomechanical modeling of an articulated skeleton actuated by skeletal muscles.
• With regard to locomotion, there are obvious differences between the hydrodynamic swimming of a fish and the natural quadrupedal gait of the cat that exploits gravity and friction due to foot-ground contact. Nevertheless, for both the fish and the cat, it is crucial to the design of the higher level behavioral modeling to abstract the locomotion ability of the animal into a set of parameterized motor controllers.
• Perceptual modeling is essential in both the cat and the fish. The perception model that we have developed can with slight modification emulate the basic visual capabilities of a cat, but one may also wish to model the auditory capability of cats.
• The artificial cat requires a behavior system at least as sophisticated at the one for the artificial fish, including habits, mental state, an intention generator and behavior routines. Certain innate characteristics, such as gender, mental variables, such as hunger, and behaviors, such as collision avoidance, are common to both fish and cats, but behaviors such as scratching in the litterbox are feline specific and would require the implementation of specific behavior routines. However, the structure of our behavior control scheme remains appropriate.

To develop more complex artificial animals patterned after humans and other primates that have sophisticated intelligent behaviors, our approach can serve in the development of a reactive behavioral substrate that supports a higher-level reasoning system.

Xiaoyuan Tu

January 1996