T. Shibata et al., HIERARCHICAL HYBRID NEUROMORPHIC CONTROL-SYSTEM FOR ROBOTIC MANIPULATORS, JSME international journal. Series C, dynamics, control, robotics, design and manufacturing, 36(1), 1993, pp. 100-109
In this paper, we present a hierarchical intelligent control system. W
e propose this system for generalization of the neural network-based c
ontroller using the higher-level control based on AI technology to acq
uire knowledge heuristically. Therefore, this system comprises two lev
els: a ''learning'' level and an ''adaptation'' level. The neural netw
orks are employed for both the long-term learning of the control proce
ss and the short-term adaptation of the dynamic process. The learning
level has a hierarchical structure for recognition and is used for the
strategic planning of robotic manipulation in conjunction with the kn
owledge base in order to expand the adaptable range to the environment
. New information from the adaptation level updates the learning level
through the long-term learning process. On the other hand, the adapta
tion is used for the adjustment of the control law to the current stat
us of the dynamic process. The motion controller at the adaptation lev
el is particularly useful in non-linear dynamical systems having uncer
tainty in the environment.