David Hanson, Stuart Baurmann, Thomas Riccio, Richard Margolin, Tim Dockins, Matthew Tavares, Kevin Carpenter
We hypothesize that well-designed intelligent characters, fusing good artistry with best-of-breed cognitive A.I., can result in greater user engagement. Realizing such designed robot characters, though, requires substantial technical innovation in both software and hardware. To simulate a character with motives and intention, a cognitive framework and authoring tools are needed, as is sensor fusion and enough machine perception, face-tracking and speech recognition to enable the character to appear aware enough to the user to forge a simulated relationship. To test the efficacy of such a Character Interface Device (CID), we developed a prototype in the form of a walking, expressive character prototype named Zeno RoboKind that brings together these systems. To facilitate the CID as a more general, intuitive form of human-machine interface device, and to bridge numerous state-of-the-art A.I. components and other cognitive architectures, we have begun to standardize an open architecture entitled Adaptive Character Engine (ACE).