On the feedforward torques and reference trajectory for flexible two-link arm

被引:14
作者
Aoustin, Y
Formal'sky, A
机构
[1] Inst Rech Cybernet Nantes, UMR 6597, F-44321 Nantes 3, France
[2] Moscow MV Lomonosov State Univ, Inst Mech, Moscow 119899, Russia
关键词
elastic vibrations; discontinuous control; continuous control; time-optimal; experimental results;
D O I
10.1023/A:1009875908235
中图分类号
O3 [力学];
学科分类号
08 ; 0801 ;
摘要
A method for synthesizing the feedforward torques and reference trajectory for a flexible two-link planar manipulator is proposed. These torques and the trajectory are not far from being time-optimal. The control law for each drive torque consists of two parts: a commanded feedforward torque and a linear angular position and velocity feedback. Initially, we theoretically prove that so-called 'fluent' control enables an individual link to reach the desired hub acceleration with little error and avoid unwanted large flexible vibrations. With this control, the dynamics of a flexible link is close to the dynamics of the associated rigid link. We begin the design of the feedforward torques and reference trajectory by computing the control torques for an arm with two rigid links which are near time-optimal control functions. These torques are discontinuous functions of time. The jumps in the control torques are not acceptable for an arm with flexible links because large elastic vibrations appear in the links. This conclusion follows from a theoretical study of individual links and also from experimentation with one-link and two-link flexible arms. Therefore, fluent control is considered here. Designing this fluent control, we transform the obtained discontinuous control into continuous functions of time, taking into account the elasticity of flexible links. During the experiments with flexible arms, these continuous functions are used as commanded feedforward torques. We also feed on-line these control torques to a mathematical model of the arm with two rigid links and compute the corresponding desired angular velocities and positions of the links as functions of time. Then we feed this reference trajectory to the linear feedback system (usual PD-controller) of the flexible arm. Designed control does not excite large elastic vibrations although, simultaneously, it is not far from being time-optimal. The designed control algorithm has been successfully implemented in the experiments.
引用
收藏
页码:241 / 265
页数:25
相关论文
共 31 条
[1]  
AKULENKO LD, 1982, PMM-J APPL MATH MEC+, V46, P465, DOI 10.1016/0021-8928(82)90031-4
[2]  
[Anonymous], 1988, MODELISATION COMMAND
[3]  
Aoustin Y., 1993, Proceedings IEEE International Conference on Robotics and Automation (Cat. No.93CH3247-4), P737, DOI 10.1109/ROBOT.1993.292233
[4]   On the synthesis of a nominal trajectory for control law of a one-link flexible arm [J].
Aoustin, Y ;
Formalsky, A .
INTERNATIONAL JOURNAL OF ROBOTICS RESEARCH, 1997, 16 (01) :36-46
[5]  
Bayo E., 1988, International Journal of Robotics & Automation, V3, P150
[6]   INVERSE DYNAMICS AND KINEMATICS OF MULTI-LINK ELASTIC ROBOTS - AN ITERATIVE FREQUENCY-DOMAIN APPROACH [J].
BAYO, E ;
PAPADOPOULOS, P ;
STUBBE, J ;
SERNA, MA .
INTERNATIONAL JOURNAL OF ROBOTICS RESEARCH, 1989, 8 (06) :49-62
[7]   A FINITE-ELEMENT APPROACH TO CONTROL THE END-POINT MOTION OF A SINGLE-LINK FLEXIBLE ROBOT [J].
BAYO, E .
JOURNAL OF ROBOTIC SYSTEMS, 1987, 4 (01) :63-75
[8]  
Berbyuk V. E., 1984, Mechanics of Solids, V19, P57
[9]  
BOOK WJ, 1985, P 2 INT S ROB RES, P203
[10]   INITIAL EXPERIMENTS ON THE ENDPOINT CONTROL OF A FLEXIBLE ONE-LINK ROBOT [J].
CANNON, RH ;
SCHMITZ, E .
INTERNATIONAL JOURNAL OF ROBOTICS RESEARCH, 1984, 3 (03) :62-75