Você está na página 1de 6

Ho-Hoon Lee

Department of Mechanical Engineering, Tulane University, New Orleans, Louisiana 70118 Phone: (504) 865-5135 Fax: (504) 865-5345 e-mail: hhlee@tulane.edu

A New Motion-Planning Scheme for Overhead Cranes With High-Speed Hoisting


This paper proposes a motion-planning method for a high-performance anti-swing control of overhead cranes, where the motion-planning problem is solved as a kinematic problem. First, an anti-swing regulating control law is proposed based on the Lyapunov stability theorem, where the proposed anti-swing control drives trolley velocity regulating error asymptotically to zero while suppressing load swing rapidly to zero for given arbitrary high-speed hoisting motions. Then a motion-planning scheme is designed based on the concept of minimumtime control, the proposed anti-swing control law, and typical antiswing crane-operation practices. The motion-planning scheme is free from the usual mathematical constraints in anti-swing control such as small swing angle, small hoisting speed, and small hoisting distance. The effectiveness of the proposed motion planning is shown by generating high-performance anti-swing trajectories with high hoisting speed and hoisting ratio. DOI: 10.1115/1.1767855

Introduction

Overhead cranes are widely used in industry to move heavy objects. However, the overhead cranes have serious problems; their motion normally induces undesirable load swing since crane loads are moved via exible ropes. Such load swing usually degrades work efciency, sometimes causing load damage and even safety accidents; accordingly, load swing should be suppressed as rapidly as possible. In practice, load swing is usually suppressed by trolley motion since overhead cranes normally have no actuators for load swing control. In addition, crane loads are usually hoisted up and down while the trolley is in motion, which frequently aggravates the load swing. In practical applications, an anti-swing trajectory of motion is divided into an accelerating zone, constant-velocity zone, and decelerating zone, and crane loads are usually hoisted up in the accelerating zone and hoisted down in the decelerating zone to avoid various obstacles on the ground. Extensive research has been carried out for the stabilization and minimization of load swing under certain constraints such as small load swing, slow hoisting speed, and small hoisting distance 116 . However, under those constraints, the existing anti-swing control laws may not be suitable for a high-performance antiswing control of overhead cranes since high-speed hoisting motions are usually involved in practical applications. High-performance motion planning is indispensable for a highperformance anti-swing control; however, only a little attention has been paid to motion planning 1,2 . Motion planning is a kinematic problem, nding certain anti-swing trolley motions for given hoisting motions without considering the required forces that cause such motions. Mita and Kanai 1 solved a minimumtime motion-planning problem without considering the hoisting motions. The resulting minimum-time control is not so effective for most applications since high-speed hoisting motions are usually required in practice for obstacle avoidance. Accordingly, a high-performance motion-planning scheme needs to be developed for a high-performance anti-swing control, especially with highspeed hoisting motions for obstacle avoidance. A major problem in the motion planning of an overhead crane
Contributed by the Dynamic Systems, Measurement, and Control Division of THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS for publication in the ASME JOURNAL OF DYNAMIC SYSTEMS, MEASUREMENT, AND CONTROL. Manuscript received by the ASME Dynamic Systems and Control Division June 24, 2002; nal revision, October 14, 2003. Associate Editor: M. Goldfarb.

is that the crane system is an underactuated system; that is, the number of control inputs of an overhead crane is less than that of degrees of freedom. An underactuated system consists of actuated dynamics and unactuated dynamics. The latter can be considered as kinematic constraints between the actuated degrees of freedom and the unactuated degrees of freedom, similar to nonholonomic constraints. Any trackable trajectory of an underactuated system should satisfy the unactuated dynamics and hence should be generated based on the unactuated dynamics. A new method of motion planning in this paper may be applied to other underactuated systems whose unactuated dynamics are marginally stable. This paper proposes a new motion-planning method for a highperformance anti-swing control of overhead cranes with highspeed load hoisting. First, an anti-swing regulating control law is proposed based on the Lyapunov stability theorem, where the proposed anti-swing control drives trolley velocity regulating error asymptotically to zero while suppressing load swing rapidly to zero for given high-speed hoisting motions. Then a new motionplanning scheme is designed based on the concept of minimumtime control, the proposed anti-swing control law, and typical anti-swing crane-operation practices. Therefore, the proposed motion-planning scheme generates typical anti-swing trajectories in practice. The proposed motion planning requires iterations for a short trolley traveling, but no iteration is required for medium and long trolley traveling. The iterations are, however, minimized by using the acceleration, deceleration, and their intervals of minimum-time anti-swing velocity proles 1 computed with reasonable rope lengths. The remainder of this paper is organized as follows. In Sec. 2, the dynamic and kinematic models of an overhead crane are described. In Sec. 3, an anti-swing regulating control law will be designed based on the Lyapunov stability theorem for given arbitrary hoisting motions. In Sec. 4, a practical high-performance motion-planning scheme will be proposed, and in Sec. 5, the proposed motion planning is evaluated by generating highperformance anti-swing trajectories. Finally, in Sec. 6, conclusions are drawn for this study.

Modeling of an Overhead Crane

Figure 1 shows the plane model of an overhead crane and its load, where x, l, and are the trolley position, hoisting rope length, and swing angle, respectively. JUNE 2004, Vol. 126 359

Journal of Dynamic Systems, Measurement, and Control Copyright 2004 by ASME

rately, the generation of hoisting trajectories should be conservative; that is, the desired rope length should be set smaller than is required at a goal position for safety. l Theorem: Suppose that r, r , r , l, and are uniformly bounded l and that l 0 for all time t 0 and 0 for all time t t 0 with l some nite positive constant t 0 . Then the following anti-swing regulating control law (5) guarantees that , , , x , x L and 0, 0 and x r asymptotically as t if 0 /2: x Kv x r 2 l cos , (5)

Fig. 1 Plane model of an overhead crane

2.1 Dynamic Model of an Overhead Crane. In this study, the mass and stiffness of the hoisting rope are neglected and the load is considered as a point mass, which is appropriate for a large class of multi-wire hoisting applications. Then the equations of motion of the crane system 3 are obtained as m x m x ml cos ml sin 2 mh fx , fh , 0, m sin d v x x 2m cos l l (1) (2) (3)

where r is a constant trolley reference velocity: the maximum velocity in the accelerating zone and zero in the decelerating zone; is a constant satisfying 0 (2g/ )( l min / max); is l a positive constant satisfying ( l ) 1.5 for all time t l 0; K v is a sufciently-large positive constant such that the bandwidth of the low-pass lter (5) is sufciently larger than the frequency content of and l. Proof of Theorem: Suppose that K v , the corner frequency of the low-pass lter 5 , is sufciently larger than the frequency content of and l. Then the anti-swing control law 5 yields the following relationship: x r 2 1 cos l sin cos2 cos , l l cos (6)

m m sin x d v h ml 2 mg cos l l ml 2 ml cos x 2mll mgl sin

x 2

(7)

Then the kinematic Eq. 4 with Eq. 7 can be rewritten as l 2 l 1 sin cos 2l l l g sin 0, (8)

where m is the load mass; m x and m h are the x traveling and h hoisting down components of the crane mass, respectively, which include the equivalent masses of the rotating parts such as motors and their drive trains; d v x and d v h denote the viscous damping coefcients associated with the x and h motions, respectively; f x and f h are the driving forces in the x and h hoisting down directions, respectively; g denotes the gravitational acceleration. 2.2 Kinematic Model for Motion Planning. The crane dynamics consists of actuated dynamics the trolley dynamics 1 and the load hoisting dynamics 2 and unactuated dynamics the load swing dynamics 3 . As stated in the introduction, the unactuated dynamics will be used for motion planning. The unactuated dynamics 3 can be rewritten as the following kinematic model for motion planning: l cos x 2l g sin 0, (4)

where 0 (2g/ )( l min / max) l guarantees g sin / l) ( l for all /2. Now, consider the following Lyapunov function candidate: V 1 2 l g 1 cos 2 l 2 l
2

0,

(9)

where 0 /2 and 0 (2g/ )( l min / max). l Take the time derivative of V along the trajectories of the kinematic system 8 : V l 1 2 g sin l 2 l 2 l 4l l l 2 l 4l l l As l 2 l 1 2 l 2 l 1 2 sin cos 2 2 l 4l l l 2 l
2 2

l l l

2 l 4l l sin cos

l 2 l 2

which describes the kinematic relationship between the trolley motion and load swing for a given hoisting motion l. In motion planning, the trolley acceleration is considered as the input and the swing angle is considered as the output. It is noted that the kinematic model 4 is independent of load mass m and hence contains no parametric uncertainties for motion planning.

3 l 2 2
2

Anti-Swing Control Under High-Speed Hoisting

sin cos

3 2 l 2

In this section, the anti-swing motion-planning problem will be solved based on the kinematic model 4 for given load-hoisting motions. Then the resulting anti-swing control law, described in the Theorem, will be used in the next section for the motion planning in the accelerating and decelerating zones. Assumption: As in practice, it is assumed that the hoisting trajectories and their time-derivatives are uniformly bounded, and that the hoisting trajectories are generated independently of the trolley motions in the accelerating, constant-velocity, and decelerating zones. It is also assumed that the goal position of the crane load is accurately known. If the goal position is not known accu360 Vol. 126, JUNE 2004

(10)

/2, sin /cos since 0 /2, and hence V is uniformly bounded. since l 0 for all time t 0 and l Therefore, V(t) dened in Eq. 9 with its time-derivative V (t) Transactions of the ASME

implies that L and /2 for all time t 0 since 0 /2. Then x L follows from Eq. 6 , x L follows from Eq. 7 , and L follows from Eq. 8 . The hoisting acceleration is uniformly bounded and satises l 0 for all time t t with some nite positive constant t . l 0 0 Therefore, integration of V (t) yields 0 2 dt , which is equivalent to L 2 . Then, as a consequence of Barbalets Lemma, 0 asymptotically as t since L , which implies that the magnitude of decreases asymptotically to zero as t . Then 0 asymptotically as t since is differentiable ( L ). Then, 0 follows from Eq. 8 . Then, from Eq. 6 , x r asymptotically as t . Remark 1: The anti-swing control law 5 allows arbitrary hoisting trajectories l, , and ) as long as they are uniformly l l bounded and l 0 and 0 for all time t t 0 with some nite l positive constant t 0 . However, in practice, the load is usually hoisted up in the accelerating zone and hoisted down in the de celerating zone. As noted in Eq. 4 , hoisting-up motions (l 0) result in negative damping and hence tend to magnify the load swing in the accelerating zone, and hoisting-down motions (l 0) lead to positive damping and hence tend to reduce the load swing in the decelerating zone.

decelerating intervals t a and t d are set to one swing period of the load. The load swing period can be determined based on the following linearized model: l0 g x, (11)

Proposed Motion Planning

In this study, a new anti-swing motion-planning scheme will be designed based on the concept of minimum-time control, typical anti-swing crane-operation practices, and the proposed anti-swing regulating control 5 , for a given hoisting trajectory l. The trolley will be accelerated as much as possible for the rst half of the accelerating zone, and then the resulting load swing will be suppressed to zero by using the proposed anti-swing control 5 while the trolley is accelerated to a normal velocity at the end of the zone. Crane load will be hoisted up in this zone for obstacle avoidance. In the constant-velocity zone, the trolley will be moving at the normal speed with zero load swing and with the load hoisted up. In this zone, additional load hoisting will not cause a load swing since 0, 0, and 0 in the beginning of the zone, as noted in Eq. 4 . However, no additional load hoisting is usually required in the constant-velocity zone. Finally, the trolley will be decelerated as much as possible for the rst half of the decelerating zone, and then the resulting load swing will be reduced to zero by using the proposed anti-swing control 5 while the trolley is decelerated to a complete stop at the end of the zone. The crane load will be hoisted down in the decelerating zone. The proposed motion planning is described in Fig. 2, where a a , a d , t a , t c , and t d denote the acceleration, deceleration, accelerating interval, constant-velocity interval, and decelerating interval, respectively. In addition, x r and r max denote the trolley-traveling distance and maximum trolley velocity, respectively; T , T , and x T are small positive constants denoting allowable upper bounds of , , and x , respectively, at the end of the accelerating and decelerating zones. Finally, TG indicates the two-different modes of trajectory generation: TG 0 for motion-planning mode and TG 1 for trajectory-generation mode. In this study, the acceleration a a , deceleration a d , accelerating interval t a , constant-velocity interval t c , and decelerating interval t d will be determined based on the velocity proles shown in Fig. 3, where the additional symbols t, r , and r max denote the time, velocity, and maximum velocity, respectively. The area under the velocity r represents the total traveling distance. When the rope length is constant, the velocity proles represent a minimum-time anti-swing velocity 1 if the accelerating and Journal of Dynamic Systems, Measurement, and Control

which was derived by linearizing the kinematic model 4 for small swing 1 with a constant rope length l 0 . Then one swing period of the load for l 0 is computed as 2 / g/l 0 seconds. According to Remark 1, the hoisting-up motions in the accelerating zone tend to magnify the load swing and the hoistingdown motions in the decelerating zone tend to reduce the load swing. Hence, in this study, the accelerating interval is set longer than the decelerating interval to reduce the load swing in the accelerating zone as follows. The initial accelerating interval t a is set to one swing period computed with the average of the average rope length and the maximum rope length in the accelerating zone, and the initial decelerating interval t d is set to one swing period computed with the average of the average rope length and the minimum rope length in the decelerating zone. For a given maximum trolley velocity, rope length, and trolleytraveling distance, the required velocity prole can be determined from Fig. 3. The anti-swing acceleration and deceleration of the trolley can be computed based on the initial accelerating and decelerating intervals. The acceleration a a in the accelerating zone is selected as a a r max /ta , and the deceleration a d in the decelerat ing zone is selected as a d r max /td . Remark 2: In the hoisting trajectory generation, triangular and/or trapezoidal velocity proles shown in Fig. 3 can be used for the hoisting-up and hoisting-down motions. Smooth velocity proles may be also used for smooth motion; for example, instead of the straight lines, cubic polynomials can be used for the velocity proles in the accelerating and decelerating zones. Remark 3: Iterations are required for a short trolley traveling with a triangular velocity prole. However, no iteration is required for a long trolley traveling with a trapezoidal velocity prole, as noted in Fig. 2. The iterations are, however, minimized by using the acceleration, deceleration, and their intervals of minimumtime anti-swing velocity proles 1 computed with reasonable rope lengths. Remark 4: In the constant-velocity zone and at goal positions, l is constant. Then the kinematic equation 8 can be rewritten as l 2 (1 sin /cos ) g sin 0 with 0. Hence, will be selected as (gl c ) 1/2 for the optimum damping of load swing, where l c denotes a constant rope length, and denotes the damping ratio. As is increased, can be decreased appropriately. For an optimal damping, a gain scheduling method can be adopted for and as a function of the minimum rope length, maximum rope length, hoisting speed, hoisting ratio, and so on. Remark 5: The proposed trajectory generation consists of an open-loop control (x a a and x a d ) and a closed-loop antiswing control 5 in series. Without loss of generality, the combination of the two different controls can be changed as follows. In the accelerating zone, set x a a for 0 t ta t a and apply Eq. 5 for t t a , with 0 ta ta 1. In the decelerating zone, set x a d for 0 t td t d and apply Eq. 5 for t td t d , with 0 td 1. Remark 6: The transition from the open-loop control minimum-time control to the anti-swing control 5 in the accelerating and decelerating zones may cause large acceleration changes. Therefore, the anti-swing acceleration at the transition should be saturated for the generation of trackable trajectories. Remark 7: At the beginning of the constant-velocity zone, there will be in general discontinuities in the swing angle and swing velocity depending on the magnitude of the allowable upper bounds T , T , and x T . However, these discontinuities normally cause no problem in real-time control since stable control laws will be used for anti-swing control. JUNE 2004, Vol. 126 361

Fig. 2 Flowchart of the proposed motion planning

Remark 8: Extension of the proposed motion-planning scheme for three-dimensional overhead cranes 3 is a little involved, but can be readily achieved.

Trajectory Generation

The proposed motion-planning scheme has been applied to the generation of high-performance anti-swing trajectories for the overhead crane shown in Fig. 1. The following parameters were used for motion planning: K v 1000, 0.75 g and 0.05 gl v for the accelerating zone, 0.45 g and 0.35 gl v for the decelerating zone, and ta td 0.5, where l v 362 Vol. 126, JUNE 2004

denotes the average rope length in the accelerating zone and in the decelerating zone. The sum of the coefcients and are 0.8 in both the accelerating and decelerating zones. The upper bounds T , T , and x T for , , and x in the accelerating and decelerating zones were all set to 1.5 10 3 rad, rad/s, and m/s, respectively. The maximum velocity r max was set to 2 m/s. As stated in Remark 6, the acceleration was saturated at 2a a in both the accelerating and decelerating zones to limit the acceleration of the trolley right after the transition from the open loop control (x a a and x a d ) to the closed-loop anti-swing control 5 . The trapezoidal rule was used for the integration of the kineTransactions of the ASME

Fig. 5 Trajectory generation for a medium traveling with load hoisting

Fig. 3 Velocity proles for trajectory references

matic model 4 and the trolley-traveling accelerations set in Sec. 4. The integration interval was chosen to be 2 ms. When the sampling interval of an applied real-time control is large, a multiple sampling for the trajectory generation or the fourth order Runge-Kutta formula 17 may be used for the accuracy of trajectory generation. As stated in Remark 3, the proposed method requires iterations for a short trolley traveling with a triangular velocity prole. However, the proposed motion planning minimizes the iterations by using the acceleration, deceleration, and their intervals of the minimum-time anti-swing velocity proles computed with reasonable rope lengths. As a result, the iterations for a short trolley traveling converged very fast, which took less than 1 second even in the worst case. Figures 4, 5, and 6 show the trajectories for medium traveling without load hoisting, medium traveling with load hoisting, and long trolley traveling with load hoisting, respectively. As shown in Figs. 5 and 6, the load is hoisted up in the accelerating zone and hoisted down in the decelerating zone in order to avoid various obstacles in the workspace. The hoisting speeds in the accelerating and decelerating zones are very high with very high hoisting ratio the ratio of the longest to the shortest rope length .

The load swing is zero in the constant-velocity zone as seen in Fig. 6 and at the end of the decelerating zones as shown in Figs. 4, 5, and 6. This shows that the proposed anti-swing control law 5 guarantees asymptotic stability and prompt damping of load swing in both the accelerating and decelerating zones. Figure 4 shows a trajectory generation with a constant rope length (l 8.5 m); for this case, the sum of the accelerating and decelerating intervals is 12 seconds two swing periods of the load 1 . In Fig. 4, the sum of the accelerating and decelerating intervals is about 12.5 seconds, which implies that the proposed motion-planning scheme generates a minimum-time anti-swing trajectory at least with a constant rope length. For the medium and long trolley traveling as shown in Figs. 5 and 6, the rope length changes from 15 m to 2 m in the accelerating zone and from 2 m back to 15 m in the decelerating zone, which covers all worst possible load hoisting in industry. The average rope length in both zones is 8.5 m. In this case, the sum of the accelerating and decelerating intervals is about 13 seconds as seen in Figs. 5 and 6, which looks quite reasonable considering the minimum-time anti-swing accelerating and decelerating intervals for a 8.5-m rope length shown in Fig. 4. As stated in Remark 1, the hoisting-up motion tends to reduce the damping and the hoisting-down motion tends to increase the damping. However, the proposed trajectory generation realized almost the same magnitude of load swing in the accelerating and decelerating zones by adjusting the accelerating and decelerating

Fig. 4 Trajectory generation for a medium traveling without load hoisting

Fig. 6 Trajectory generation for a long traveling with load hoisting

Journal of Dynamic Systems, Measurement, and Control

JUNE 2004, Vol. 126 363

intervals, as described in Sec. 4. The damping in the accelerating and decelerating zones can be readily increased; however, additional damping causes sluggish over-damped motion. A higherspeed load hoisting requires a larger trolley acceleration, as noted in Figs. 4 and 5. A large normal trolley velocity results in a large trolley acceleration, as seen in Figs. 5 and 6. As stated in Remark 6, the anti-swing control 5 causes sudden changes in trolley acceleration to reduce the load swing at the transition from the open-loop control to the closed loop control, especially in the accelerating zone, as noted in Figs. 4, 5, and 6. In this way, the proposed anti-swing motion-planning scheme prevents load swing from growing large, which is a favorable feature of the proposed motion planning for safe crane operation. Finally, the values of and used in this simulation guarantee the same trajectory-generation performance for all practical combinations of traveling distance, maximum traveling velocity, hoisting distance, and hoisting speed as long as the maximum hoisting ratio is kept not greater than 15/2. When the maximum hoisting ratio increases, the values of and need to be retuned.

quently, it can be concluded that the proposed approach has a great potential for high-efciency anti-swing control of overhead cranes.

References
1 Mita, T., and Kanai, T., 1979, Optimal Control of the Crane System Using the Maximum Speed of the Trolley in Japanese With English Abstract , Trans. Soc. Instrument. Control Eng. Japan , 15, pp. 833 838. 2 Lee, H.-H., 2002, A Path-Planning Strategy for Overhead Cranes With High Hoisting Speed, Proc. of IMECE 2002, Paper No. IMECE2002-DSC-33127, New Orleans, Louisiana. 3 Lee, H.-H., 1998, Modeling and Control of a Three-Dimensional Overhead Crane, ASME J. Dyn. Syst., Meas., Control, 120, pp. 471 476. 4 Lee, H.-H., Cho, S.-K., and Cho, J.-S, 1997, A New Anti-Swing Control of Overhead Cranes, Proc. of IFAC Int. Workshop on Automation in the Steel Industry, Pohang Korea, pp. 137142. 5 Auernig, J. W., and Troger, H., 1987, Time-Optimal Control of Container Cranes With Hoisting of the Load, Automatica, 23, pp. 437 446. 6 dAndrea-Novel, B., and Boustany, F., 1991, Adaptive Control of a Class of Mechanical Systems Using Liniarization and Lyapunov Methods. A Comparative Study on the Overhead Crane Example, Proc. of the 30th IEEE Conf. on Decision and Control, Brighton, England, Dec., pp. 120125. 7 Moustafa, A. F., and Ebeid, A. M., 1988, Nonlinear Modeling and Control of Overhead Crane Load Sway, ASME J. Dyn. Syst., Meas., Control, 110, pp. 266 271. 8 Ohnishi, E., Tsuboi, I., and Egusa, T., 1981, Automatic Control of an Overhead Crane, IFAC World Congress, Kyoto, Japan, pp. 18851890. 9 Parker, G. G., Petterson, B., Dohrman, C., and Robinett, R. D., 1995, Command Shaping for Residual Vibration Free Crane Maneuver, Proc. of American Control Conference, Seattle, pp. 934 938. 10 Piazzi, A., and Visioli, A., 2002, Optimal Dynamic Inversion Based Control of an Overhead Crane, IEE ProceedingsControl Theory and Applications, 149, pp. 405 411. 11 Ridout, A. J., 1987, New Feedback Control System for Overhead Cranes, Electric Energy Conference, Adelaide, pp. 135140. 12 Sakawa, Y., and Sano, H., 1997, Nonlinear Model and Linear Robust Control of Overhead Traveling Cranes, Nonlinear Anal.: Real World Appl., 304, pp. 21972207. 13 Sakawa, Y., and Sindo, H., 1982, Optimum Control of Container Cranes, Automatica, 18, pp. 257266. 14 Singhose, W., Porter, L., Kenison, M., and Kriikku, E., 2000, Effects of Hoisting on the Input Shaping Control of Gantry Cranes, Control Eng. Pract., 8, pp. 11591165. 15 Strip, D. R., 1989, Swing-Free Transport of Suspended Objects: A General Treatment, IEEE Trans. Rob. Autom., 5 2 , pp. 234 236. 16 Yu, J., Lewis, F. L., and Huang, T., 1995, Nonlinear Feedback Control of a Gantry Crane, Proc. of American Control Conference, Seattle, pp. 4310 4315. 17 Press, W., Flannery, B., Teukolsky, S., and Vetterling, W., 1986, Numerical Recipes: The Art of Scientic Computing, Cambridge University Press.

Conclusion

In this paper, a new anti-swing motion-planning method has been proposed for a high-performance anti-swing control of overhead cranes with high hoisting speed and large hoisting distance. The proposed motion-planning scheme has been designed based on the proposed anti-swing control law, the concept of minimumtime control, and typical anti-swing crane-operation practices. The stability of the anti-swing control law has been proven based on the Lyapunov stability theorem and has been also shown by computer simulation. The damping of the anti-swing control law can be directly controlled using damping gains and has been also shown to be excellent in the trajectory generation. In addition, the anti-swing control law allows any pattern of load hoisting in the accelerating and decelerating zones, and drives the trolley velocity regulating error asymptotically to zero. The trajectory generation for various cases shows that the proposed motion-planning scheme generates typical anti-swing trajectories in practice. Furthermore, the kinematic model does not include any disturbances and uncertainties for motion planning, and hence the proposed motion-planning scheme by nature does not have any uncertainties and robustness problems. Conse-

364 Vol. 126, JUNE 2004

Transactions of the ASME

Você também pode gostar