Knowledge-based control approach for robotic manipulators

Abstract

The integration of hard algorithmic control and soft knowledge-based control is addressed, in the specific context of robotic manipulator control. In particular, a hierarchical control structure having a high-speed hard controller at the lowest level and an intelligent observer and an intelligent tuner at the upper levels, is discussed. The rationale for a specific control structure is explored, and the necessary qualitative and analytical foundation for the control structure is presented. To demonstrate the development and feasibility of this control structure, an example application has been programmed on a SUN workstation. A two degree-of-freedom robot was simulated (a C program) as a separate UNIX process. The intelligent observer was developed using an available AI toolkit. The top-level intelligent tuner was developed from a valid set of linguistic tuning rules for proportional-integral-derivative (PID) servos, using fuzzy system concepts, and implemented as a set of decision tables. The control structure is evaluated on the basis of the simulation results.