L. Whitcomb, A. Rizzi, and D. Koditschek. Comparative experiments with a new adaptive controller for robot arms. IEEE Trans. Robotics and Automation, 9(1):59--70, Feb 1993. 16  Home/Search   Document Details and Download   Summary   Related Articles   Check

....video file in JPEG format demonstrating the arm and these experiments is available free via anonymous fp from alpha.me.jhu.edu, filename Ibrce.jpg . Please contact the first author (11w jhu.edu) with questions. The Effect of Model Based Rigid Body Feedfro ward Recent experimental studies, e.g. [27], have shown modelbased control algorithms for trajectory tracking to offer performancc superior to their non model based counterparts over a wide rmge of operating conditions. This is in contrast to several early experimental studies of model based robot trajectory tracking algorithms which ....

....errors rs. time (middle) Actual and reference tool tip surface normal force vs. time (bottom) Tool tip speed = O. 0628 meters second. steady state performance. These start up transients are typical of adaptive control systems for which the adaptive parameter values ar initialized to zero [27]. The top and middle graphs show the steady state force tracking error under the IDCFA controller to be under 1 centimeter, while that for PDF is over 4 cm. The PDF posifipn tracking error is four times worse than for IDCFA. The bottom graphs show the actual and reference tool tip surface normal ....

[Article contains additional citation context not shown here]

L.L. writcomb, A. Rizzi, and D.E. Koditschek, "Comparative Experiments with a New Adaptive Controller for Robot Arms," in IEEE Transac- tions on Robotics and Automation, 9(1):59 70, 1993.

....e.g. 70] 82] In fact, throughout the field of robotics, there is a general tendency to shy away from defining specific task performance metrics, particularly as tasks become complex. 21 Although attempts have been made to introduce uniform quantitative metrics to specific subfields (e.g. [107]) this type of analysis has not taken hold, particularly in the field of autonomous mobile robotics. This observation is evident from the number of papers that use a diagram showing the path that their robot(s) took on a particular trial (i.e. validation ) as proof that their algorithm works, ....

L. L. Whitcomb, A. A. Rizzi, and D. E. Koditschek. Comparative experiments with a new adaptive controller for robot arms. IEEE Transactions on Robotics and Automation, 9(1):59--70, February 1993.

....as is standard in the robotics community. However, sweeping the issue of dynamics under the carpet requires the use of sophisticated robot control techniques that require precise parametric knowledge of the robot s kinematics and dynamics, such as adaptive inverse dynamics based control strategies [22]. This extra complexity seems superfluous given the simple endpoint convergence objective of most visual servoing algorithms. In other words, the specific reference trajectories generated by kinematic controllers are merely a means to achieve an end point task; tracking those trajectories exactly ....

Louis L. Whitcomb, Alfred A. Rizzi, and Daniel E. Koditschek. Comparative experiments with a new adaptive contoller for robot arms. IEEE Transactions on Robotics and Automation, 9(1):59--70, Feb 1993.

....9. Robot tracking a unit circular contour under PVFC. nonadaptive model based timed trajectory control law, known as the desired compensation control law (DCCL) 5] The adaptive version of the DCCL has been shown to perform superiorly to other adaptive model based trajectory tracking controllers [21]. The desired circular contour was centered at the origin of the joint coordinate space and had a radius of 1. The desired velocity field was designed so that if , then the contour would be followed at an angular contouring velocity of 1 rad s. In all experiments, joint friction was partially ....

L. L. Whitcomb, A. A. Rizzi, and D. E. Koditschek, "Comparative experiments with a new adaptive controller for robot arms," IEEE Trans. Robot. Automat., vol. 9, pp. 59--70, Feb. 1993.

.... note that it may be equivalently written in the form = W (q 1 ; q 2 ; 3 ; 4) where 3 is a vector of dynamical parameters (that we still presume to be known at this point in the discussion) and W is an array of known nonlinear functions that depend upon the robot s known kinematics[1, 42]. This control law results in the nonautonomous nonlinear error dynamics e 1 = e 2 e 2 = 0M 01 (q 1 ) C(q 1 ; q 2 )e 2 Ke] 5) As in the case of the computed torque algorithm (2) this error system is globally asymptotically stable when K = K T 0. However, the demonstration is no ....

....ran and computed the error norm average and standard deviation over ten runs with identical controllers, plants, and reference trajectories for each of the listed controllers. The standard deviation of the error norm over the ten runs was observed to be typically less than one percent of their mean[42]. 2 erformance ene ts ue to the Adaptive Algorithms Figure 3 shows PD position tracking error norm ( L 2 [e(t) e(t) e 1 (t) e 2 (t) e 3 (t) T ) of between 13 and 25 , IDR errors of 3 to 6 , and IDRA errors of roughly 2 over a range of reference trajectories. The figure 6 shows the ....

[Article contains additional citation context not shown here]

Louis L. Whitcomb, Alfred A. Rizzi, and Daniel E. Koditschek. Comparative experiments with a new adaptive controller for robot arms. Technical Report 9101, Yale University, Center for Systems Science, New Haven, Connecticut, USA, January 1991. 19

....positive constant ffl 0 , independent of the initial conditions , such that the equilibrium [x T p ; f; x T p ; f ; T ] T = x T pd ; f d ; 0; 0; g T (x d ) T is GAS. P roof . See section 4.1. Remark 3.1. As will become clear in the proof, we have used a normalized ffl (like in (Whitcomb et al. 1991), Kelly 1993a) to dominate in the Lyapunov derivative a cubic term. Namely, the term fflk c k xkk xk 2 in (23) Very recently, Arimoto 1994a) showed that by feeding back a saturated position error in a PID control law, it is also possible to dominate this cubic term thus achieving GAS. ....

Whitcomb, L., A. Rizzi and D. Koditscheck (1991). Comparative experiments with a new adaptive controller for robot arms. In `Proc. IEEE Conf. Robotics Automat.'. Sacramento, CA. pp. 2--7.

....v2 bwd desired v2 t = 20 = 8 t = 12 Figure 8 goes here. File = velocity.eps Fig. 8. q 2 =j j where 2 = k( q; q) E, and V 2 Law (DCCL) 5] The adaptive version of the DCCL has been shown to perform superiorly to other adaptive model based trajectory tracking controllers [21]. The desired circular contour was centered at the origin of the joint coordinate space and had a radius of 1. The desired velocity eld was designed so that if q = V( q(t) then the contour would be followed at an angular contouring velocity of 1 rad s. In all experiments, joint friction was ....

L. L. Whitcomb, Alfred A. Rizzi, and Daniel E. Koditschek, \Comparative experiments with a new adaptive controller for robot arms," IEEE Transaction on Robotics and Automation, vol. 9, no. 1, pp. 59-70, feb 1993.

.... methods can be applied, like nonlinear decoupling control [7] resolved acceleration control [12] or the computed torque method [2, 15] These approaches depend on the precise knowledge of the robot parameters and of the possibly varying load, and perform poorly when the model is not accurate [2, 18]. We focus on the case where the actuator can provide precise driving torque without the need for torque sensor feedback. Now joint torque measurements can be used for positive feedback to compensate the effects of the manipulator dynamics. For this purpose we have designed and built a new torque ....

L. Whitcomb, A. Rizzi, and D. Koditschek. Comparative experiments with a new adaptive controller for robot arms. IEEE Trans. Robotics and Automation, 9(1):59--70, Feb 1993.

....limited relevance whenever an exact measurement of this quantity is not available. Well known solutions to this problem rely on model based adaptive schemes. Three examples are the composite adaptive controller in [88] the passivity based controller in [4] and the indirect adaptive controller in [104]. In what follows, we sketch the basic common idea behind these treatments. The key simplifying assumption is that the unknown parameters enter linearly the Euler Lagrange equations. In particular, assume that the inertia tensor M(q) satisfies M(q) P i i M i (q) where i 2 R are unknown ....

L. L. Whitcomb, A. A. Rizzi, and D. E. Koditschek. Comparative experiments with a new adaptive controller for robot arms. IEEE Transactions on Automatic Control, 9(1):59--70, 1993.

No context found.

Louis L. Whitcomb, Alfred A. Rizzi, and Daniel E. Koditschek, "Comparative experiments with a new adaptive contoller for robot arms," IEEE Transactions on Robotics and Automation, vol. 9, no. 1, pp. 59--70, Feb 1993.

No context found.

L. L. Whitcomb, A. Rizzi, and D. E. Koditschek, "Comparative experiments with a new adaptive controller for robot arms," IEEE Trans. Robot. Automat., vol. 9, pp. 59--70, Feb. 1993.

....of exact model based tracking of robot arms without exact linearization was first reported in [53] 54] along with an adaptive extension that is globally asymptotically stable in tracking error and globally stable in plant parameter error. Variations of this general approach are reported in [52] [63]. The application of this approach to the control of underwater vehicles is the nonlinear controller (NL) outlined in Section III C, the adaptive nonlinear controller (ANL) outlined in Section III E, and the epsilon adaptive nonlinear controller (#ANL) of Section III F. For the special problem ....

....6. Using Corollary 2.9 in [51] with points 3 and 5, ### ### #######. In summary we conclude that all signals remain bounded, and the velocity error asymptotically tracks to zero. This approach is an application to underwater vehicles of the general methodology originally reported in [53] 54] [63]. D. AEL: Adaptive Exact Linearizing Model Based Controller The adaptive exact linearizing model based controller (AEL) for a single degree of freedom takes the form ## #### # ##### # #### # # ## #### # ### # # # ## # # ## (18) # #### are adaptive estimates of the scalar plant parameter ....

[Article contains additional citation context not shown here]

L. L. Whitcomb, A. Rizzi, and D. E. Koditschek. Comparative experiments with a new adaptive controller for robot arms. IEEE Transactions on Robotics and Automation, 9(1):59--70, February 1993.

....experimental data on the comparative performance of the IDCF (13) IDCFA (15) and PDF (11) force control algorithms. 1) The Effect of Model Based Rigid Body Feedforward: Given the diversity of experimental results for model based control of unconstrained robot motion, e.g. 1] 8] 14] 23] [25], 27] in the present context of force control it seems essential to investigate the comparative performance of model based versus nonmodel based force controllers. The purpose of this section is to compare the performance of the nonmodel based PDF controller (11) with the fully adaptive ....

....was 0.628 m s. These graphs show the IDCFA controller to exhibit a start up tracking transient which converge after a few seconds to near steady state performance. These start up transients are typical of adaptive control systems for which the adaptive parameter values are initialized to zero [25]. The PDF position tracking error is approximately four times worse than for IDCFA. The force tracking performance of both controllers is poor. The PDF steady state force errors are about two times worse than for IDCFA. We conclude that IDCFA offers tracking high speed performance superior to ....

[Article contains additional citation context not shown here]

L. L. Whitcomb, A. Rizzi, and D. E. Koditschek, "Comparative experiments with a new adaptive controller for robot arms," IEEE Trans. Robot. Automat., vol. 9, pp. 59--70, Feb. 1993.

....the proposed applications include: i) precise joint level torque control, ii) high structural vibration frequencies, iii) high end point stiffness, iv) low weight, and (v) high static strength. Previous reports addressing the design and control of high performance robot arms, e.g. IS] 15] [21], 3] 5] have established that the arm structural elasticity and the torque ripple of permanent magnet motors are the two main causes that degrade the fidelity of joint level torque control and cause oscillations in the position and force control loops. For example, a comprehensive experimental ....

....results for the new arm and compares the experimental results to those predicted by the FEM simulations. Section V presents experimental data that quantifies the tracking performance of the arm. II. KINEMATIC CANDIDATE SELECTION Previous results on robots having geared reducers, e.g. 8] 15] [21], and direct drive robots, e.g. 1] 14] 18] 6] 7] 4] 20] show that semi direct drive actuation for the arm s distal links, incorporating a mechanical linkage for power transmission, offers the best practical compromise between the two for high performance applications. On reviewing ....

[Article contains additional citation context not shown here]

L. L. Whircomb, A. Rizzi, and D. E. Koditschek, "Compara- tive experiments with a new adaptive controller for robot arms," IEEE Transactions on Robotics and Automation, vol. 9, no. 1,

....tion that A(t) 6 ;2 in (6) for the case of a known environ ment compliance. However, we do not lose any generality by making this additional restriction since most velocity controllers which guarantee asymptotically exact velocity tracking also guarantee that the velocity error is ;2, e.g. 22] [32]. 3. A Useful Lemma In this section we state a lemma that will be used later in the paper. We note that the result presented in the lemma is a minor generalization of the well known result from systems theory that the state of an asymptotically stable LTI system asymptotically decays to zero ....

L. L. Whircomb, A. Rizzi, and D. E. Koditschek. Comparative experiments with a new adaptive controller for robot arms. IEEE Transactions on Robotics and Automation, 9(1):59-70, February 1993.

....B. Inner Loop Velocity Control: Assumptions In addition to traditional proportional (P) and proportional integral (PI) velocity control schemes (which guarantee bounded velocity errors for bounded velocity trajectories) a variety of model based position velocity controllers (e.g. 29] [30]) have been reported in robot position control literature. These controllers guarantee asymptotically exact position and velocity tracking for all position trajectories for noncontact motion of the robot. For example, the 1 D robot plant (4) under the velocity control law (5) exhibits ....

....inner loop velocity tracking, we require an additional assumption that . We do not lose significant generality by making this additional restriction since most velocity controllers which guarantee asymptotically exact velocity tracking also guarantee that the velocity error is , e.g. 29] [30]. III. A USEFUL LEMMA In this section, we state and prove a lemma that will be used later in the paper. We note that the result presented in the lemma is a minor generalization of the well known result from systems theory that the state of an asymptotically stable LTI system asymptotically ....

L. L. Whitcomb, A. Rizzi, and D. E. Koditschek, "Comparative experiments with a new adaptive controller for robot arms," IEEE Trans. Robot. Automat., vol. 9, pp. 59--70, Feb. 1993.

....control of robot arms without exact linearization was first reported in [36, 37] along with an adaptive extension that is globally asymptotically stable in position and velocity tracking error and globally stable in plant parameter error. Variations of this general approach are reported in [35, 42]. The application of this approach to the control of underwater vehicles is the nonlinear controller (NL) outlined in Section 3.3, the adaptive nonlinear controller (ANL) outlined in Section 3.5, and the epsilon adaptive nonlinear controller (#ANL) outlined in Section 3.6. These ideas have each ....

....to the Lyapunov function. If # is su#ciently small, the Lyapunov function remains positive definite and radially unbounded and its time derivative is locally negative definite. This local stability result was first reported independently in [4, 28, 41] A globally stable variation was reported in [42]. Applying this approach to the present plant, 3) the control law (27) remains unchanged, resulting in the closed loop dynamical system (28) # m#x(t)#v(t) 33) 0.5##(t) # 1 ##(t) For # su#ciently small, this function is positive definite and radially unbounded. Choosing the new ....

[Article contains additional citation context not shown here]

L. L. Whitcomb, A. Rizzi, and D. E. Koditschek. Comparative experiments with a new adaptive controller for robot arms. IEEE Transactions on Robotics and Automation, 9(1):59--70, February 1993.

....to zero at the beginning of each run. Coulomb and viscous friction model terms were included in the implementation of IDCF and IDCFA, though omitted from the derivation of Section 2 for clarity [26] 3.1. The e#ect of model based rigid body feedforward Recent experimental studies, e.g. [27], have shown modelbased control algorithms for trajectory tracking to o#er performance superior to their non model based counterparts over a wide range of operating conditions. This is in contrast to several early experimental studies of model based robot trajectory tracking algorithms which ....

....0.628 meters second. These graphs show the IDCFA controller to exhibit start up tracking transients which converge after a few seconds to near steady state performance. These startup transients are typical of adaptive control systems for which the adaptive parameter values are initialized to zero [27]. The top and middle graphs show the steady state force tracking error under the IDCFA controller to be under 1 centimeter, while that for PDF is over 4 cm. The PDF position tracking error is four times worse than for IDCFA. The bottom graphs show the actual and reference tooltip surface normal ....

[Article contains additional citation context not shown here]

L. L. Whitcomb, A. Rizzi, and D. E. Koditschek. Comparative experiments with a new adaptive controller for robot arms. IEEE Transactions on Robotics and Automation, 9(1):59--70, 1993.

....force control algorithms implemented on the new robot system, and reports data from comparative force control experiments. This examination of several adaptive and non adaptive force controllers was also intended determine if our earlier experience with non contact adaptive model base control, [35], would be corroborated in the present context of constrained hybrid control. The data shows the new model based force controllers to outperform their non model based counterpart. To the best of our knowledge, these experiments compare the performance of di#ering force controllers over a wider ....

....presents actual experimental data on the comparative performance of the IDCF (13) IDCFA (14) and PDF (11) force control algorithms. 4.2. 1 The E#ect of Model Based Rigid body Feedforward Given the diversity of experimental results for model based control of unconstrained robot motion, e.g. [1, 9, 14, 17, 35, 36, 32], in the present context of force control it seems essential to investigate the comparative performance of model based versus non model based force controllers. The purpose of this section is to compare the performance of the non model based PDF controller (11) with the fully adaptive model based ....

[Article contains additional citation context not shown here]

L. L. Whitcomb, A. Rizzi, and D. E. Koditschek. Comparative experiments with a new adaptive controller for robot arms. IEEE Transactions on Robotics and Automation, 9(1):59--70, 1993.

....micromanipulation concept as applied to retinal microsurgery 2 Force Control for Steady Hand Manipulation There is a large body of literature concerning provably stable control techniques for robots. Standard paradigms include 1) pre programmed trajectory control of position [4, 25] and force [40, 41]; 2) fully autonomous robots (e.g. 20, Preprint of a paper to appear in the Proceedings of the 2000 IEEE International Conference on Robotics and Automation 34] and 3) master slave teleoperators (e.g. 12, 24, 42] In our case, we are interested stable control methodologies for cases where ....

L. L. Whitcomb, A. Rizzi, and D. E. Koditschek, "Comparative experiments with a new adaptive controller for robot arms," IEEE Transactions on Robotics and Automation, (9) 1:59-70, 1993.

....they are still allowed to directly manipulate the instruments. 2.1 Shared Autonomy and Cooperative Control There is a large body of literature concerning provably stable control techniques for robots. Standard paradigms include 1) pre programmed trajectory control of position [13, 14] and force [15, 16]; 2) fully autonomous robots (e.g. 17 19] and 3) master slave teleoperators (e.g. 20 22] In our case, we are interested in developing provably stable controls for cases where both the robot and the human manipulate a single tool in contact with a compliant environment. The work most ....

Whitcomb, L.L., A. Rizzi, and D.E. Koditschek, Comparative experiments with a new adaptive controller for robot arms. IEEE Transactions on Robotics and Automation, 1993. 9(1): p. 59-70.

....they are still allowed to directly manipulate the instruments. 2.1 Shared Autonomy and Cooperative Control There is a large body of literature concerning provably stable control techniques for robots. Standard paradigms include 1) pre programmed trajectory control of position [13, 14] and force [15, 16]; 2) fully autonomous robots (e.g. 17 19] and 3) masterslave teleoperators (e.g. 20 22] In our case, we are interested in developing provably stable controls for cases where both the robot and the human manipulate a single tool in contact with a compliant environment. The work most relevant ....

Whitcomb, L.L., A. Rizzi, and D.E. Koditschek, Comparative experiments with a new adaptive controller for robot arms. IEEE Transactions on Robotics and Automation, 1993. 9(1): p. 59-70.

....they are still allowed to directly manipulate the instruments. 2.1 Shared Autonomy and Cooperative Control There is a large body of literature concerning provably stable control techniques for robots. Standard paradigms include 1) pre programmed trajectory control of position [13, 14] and force [15, 16]; 2) fully autonomous robots (e.g. 17 19] and 3) masterslave teleoperators (e.g. 20 22] In our case, we are interested in developing provably stable controls for cases where both the robot and the human manipulate a single tool in contact with a compliant environment. The work most relevant ....

Whitcomb, L.L., A. Rizzi, and D.E. Koditschek, Comparative experiments with a new adaptive controller for robot arms. IEEE Transactions on Robotics and Automation, 1993. 9(1): p. 59-70.

No context found.

L. Whitcomb, A. Rizzi, and D. Koditschek. Comparative experiments with a new adaptive controller for robot arms. IEEE Trans. Robotics and Automation, 9(1):59--70, Feb 1993. 16

No context found.

L. L. Whitcomb, A. A. Rizzi, and D. E. Koditschek, "Comparative experiments with a new adaptive controller for robot arms," IEEE Transactions on Robotics and Automation,vol. 9, no. 1, pp. 59--70, 1993. 124 Appendix: List of symbols related to the acuatator.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC