Adaptive Control Algorithms

Abstract

In this chapter we shall present the theoretical background of various adaptive robot control algorithms. This subject is relatively new in robotic research. Within the past six years, only a handful of people have been actively working on this subject. Nevertheless, this area will surely be one of the most interesting in the nearest future, because the classical controllers cannot always satisfy the stability conditions, even if designed to be robust with respect to parametric and state disturbances (see Chapter 2). Thus, the adaptive control algorithm should be considered as an up-grade over the classical control approach. The adaptive control algorithms are often much more complex in numerical sense than nonadaptive laws. Also, it is more difficult to prove the stability of the overall system. But, the adaptive controllers offer more opportunities, especially when the robot works in ambient conditions which are not completely known in advance. This is the case, for example, when the payload mass is not predefined. Of course, this does not mean that the robot controllers should always be designed to be adaptive. On the contrary, in our opinion, the adaptive algorithms should be employed only when simple classical controllers cannot achieve desirable performances.