Processor Performance Required for Decentralized Kinematic Control Algorithm of Module-Type Hyper-Redundant Manipulator

Abstract

Based on the great progress of studies on decentralized autonomous systems, the decentralized autonomous control mechanism has expand its territory of applications to the control of redundant manipulators. Because local control systems are closely related in the decentralized autonomous control of redundant manipulators, the performance of local control systems requires careful consideration, particularly in treating the dynamic aspects of manipulators. In this paper, computer simulation is used to assess the effects of the processor performance of a decentralized autonomous control algorithm in adaptation to the reactive motion caused by the failure of actuators under microgravity conditions. The results show that the algorithm is so simple that conservative 16-bit processors are feasible as the local processor of the decentralized autonomous control algorithm. These results suggest that the decentralized autonomous control algorithm may have great advantages in hardware architecture cost and resources. The effect of conflicts among local processors on system performance is also discussed.

Adaptability of decentralized kinematic control algorithm to reactive motion under microgravity

“…3: Task achivability when first joint fails joint. This has been reported in previous work [9]. Figure 2 illustrates the typical motion pattern in a few calculation cycles when the first joint failed.…”

“…Since the difficulty of the adaptation increases according to the decrement of the friction coefficient, the a level of success that can be achived monotonicaIly decreases according to the decrement of the friction coefficient. The slightly increment in the achived level of success, which is discussed as the effects of "confrict" between local processors in previous work [9], is also seen in this condition. The trace of the achived level of success can be divided into two groups.…”

“…That is, when the reactive motion is not so great in the first step, the control system can converge the manipulator motion and succeed in the positioning task. The interval of switching from strategy to strategy is much shorter than the manipulator motion when we consider the feasible calculation cycle [9]. This is supported by the fact that there is no apparent "easy" condition for the third joint, because the errors in the posture of the first step are larger when the third joint fails than when either the first joint fails or the fifth joint fails.…”

“…We have reported that to a certain extent, this algorithm can autonomously adapt to the reactive motion which is caused by failure of the actuator under microgravity conditions [7-91. It is suggested that the adaptability in such dynamic phenomena depends highly on the performance of the local control processor [9].…”