文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > 高等教育 > 工学 > A direct adaptive control scheme for under-actuated dynamic
侵权投诉

A direct adaptive control scheme for under-actuated dynamic

文献

Procmdlnga of the 32nd Contrrmcw

on DUl8lOn and Control San Anionlo, Taxas Dmmbor 1993

TA5 11:lO

-

A Direct Adaptive Control Scheme for Under-Actuated Dynamic Systems You-Liang Gu Department of Electrical and Systems Engineering Oakland University Rochester, Michigan 48309-4401 Abstract This paper presents a class of under-actuated dynamic systems and the adaptive control methodology. A system is said t o be under-actuated if it has a t least one passive joint (i.e. not driven by actuators). The under-actuated dynamic systems have found many meaningful applications in either space explorations or advanced manufacturing operations. Due t o the passive joints, input channels become deficient, and dynamic characteristics fundamentally differ from a full-actuated system. The dynamic parameter-linearity is destroyed by the under-actuated nature. To overcome the difficulty, an extended dynamic model and a normal-form augmentation approach are proposed. Using the normal-form augmentation approach, the parameter-linearity can be recovered, and a direct adaptive control scheme can be developed t o stabilize the system against parameter uncertainty. and space applications. First, the space robot systems have played important roles in space operations. A space robot with a partially or fully passive base has been justified as an upper-actuated dynamic system. The research on space robot control with task planning is critically demanded for optimal space manipulation. Second, in today’s advanced manufacturing applications, robot manipulators are challenged to provide more flexible, faster and more powerful operations. An industrial robot conventionally carries a single-link tool to execute a task. If the tool is required t o move faster, the robot motion control has to be speeded up, resulting in lower repeatability and lower accuracy. Can we keep the robot motion relatively slow while the tool moves fast? The martial-arts superstar Bruce Lee gave us an interesting inspiration. He used to play a pair of numbchucks as his favorite weapon. By observation, each numbchuck consists of two flexible but passive joints that link three hardwood segments together. It can be seen that when he moved the bottom segment, the top one can produce a fast speed so that a tremendous amount of impact may act on a targeting object, although the mass of each segment is light. This shows that a robot tool may be imitated t o have one or more passive joints for achieving higher tool speed and stronger impact without a need of fast robot body motion. Recently, for full-actuated robots, a direct adaptive control approach has been developed to effectively compensate tracking errors due t o parameter uncertainty[1,2]. This direct adaptive control approach is based on the fact that all dynamic parameters appearing in the robot dynamic equation are linear, i.e., a full-actuated robot system has a linear parameterization property. However, this property no longer holds by an under-

actuated dynamic system. The main reason is due t o the non-holonomic constraints involved in the dynamic relation, or alternatively, due t o the existence of nontrivial internal dynamics. Therefore, we are unable to design a parameter adaptation law associated with a nonlinear feedback control scheme by using the same way as that for a full-actuated system. In this paper, we will study a new effective dynamic model for both upper- and lower-actuated systems to overcome the aforementioned obstacle in their adaptive control design. This leads to the development of an extended dynamic model and a normal form augmentation approach to the con-

1. Introduction A multi-axis dynamic system, such as a robot system, is often driven by actuators, the number of which is equal to the number of motion axes. We call such a fully driven machine the full-actuated dynamic system. For instance, a six-joint industrial robot has six motors which maneuver the robot arm to generate six degrees of freedom (d.0.f.) motion in 3D working space. If one (or more) motor is taken off and the joint is freely released, then the six-joint robot becomes an under-actuated dynamic system. The wellknown inverted-pendulum (or called a cart-pole system) is one of the simple but typical two-joint under-actuated systems, where only the cart is driven by a linear force while the inverted pendulum is mounted on the cart for free rotation. A space robot system is another realistic example, in which the robot base (either a space station or a space vehicle) is floating in free space if the base is passive (i.e. no attitude or position control). Based on the location of free-joints, we categorize all under-actuated dynamic systems into three major classes: the upper-actuated, the lower-actuated and the mixed systems. It can be seen that the space robot system with a floating base belongs to the upper-actuated category, while the inverted-pendulum system is of the lower-actuated type. The study of under-actuated dynamic systems and their control methodology has significant impact on both industrial

1

0191-2216/93/$3.00 1993 IEEE 0

1625

A direct adaptive control scheme for under-actuated dynamic

第1页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email