Biologically Inspired Robotics

Humanoid robots or "animates" are capturing the imagination. It is pure faustian drive that will help them into existence. Potential economic applications are serving more as an excuse than a purpose. Animates will be formed in the conceptual mold we have long developed around animals and humans.

Existing industrial robots are built in narrowly algorithmic style and can function only in a totally predictable environment with pre-specified goals. (This is actually their virtue: they make economic sense not as cheap labor but in terms of precision, speed and reliability.)

Letting autonomously acting physical agents loose on uncontrolled environments is an enormous challenge. Their behavior must be flexible and robust, they must be situation-aware to plan their action, they must establish and recognize sensible goals, perceive the environment, control the motion of their hands and "legs" to navigate and manipulate, and they must be physically and energetically fit. All these functions are to be integrated into one compact, self-reliant system.

Animates will serve as important pacemaker for organic computing. Whereas AI, or computers in general, live in a formal, encoded universe, always having human intelligence and perception as mediator to the environment, animates must cope with the world directly. Therefore "embodiment", the direct confrontation of a system with the physical world through concrete hardware, is accepted by the robotics community as both challenging and creative condition. There is no strong economic incentive to develop animates, at least for awhile. Correspondingly, they must be developed on shoe-string budgets in academic and industrial labs. Classical methods of software development would be too expensive by orders of magnitude. Without having to fear their competition, organic computing will have space to expand into. Self-organization, evolution, adaptation and learning will have to supplant the algorithmic style.