Hierarchical Hybrid Control of Manipulators: Artificial Intelligence in Large Scale Integrated Circuits

摘要

A system that must regulate many variables cannot have individual recipes for responding to each possible combination of variables. Both in practical engineering and in control of muscular systems, low-level subsystems automatically provide crude approximations to the proper response. Through low-level tuning of these approximations, the proper response variant can emerge from standardized high-level commands. Such systems are expressly suited to emerging large-scale integrated circuit technology. The authors have formulated mathematical theory that reveals significant informational units in this style of control, and are developing software for realizing such information structures. (Author)