William C. Messner

A new adaptive learning rule

A method is presented for nonlinear function identification and application to learning control. The control objective is to identify and compensate for a nonlinear disturbance function. The nonlinear disturbance function is represented as an integral of a predefined kernel function multiplied by an unknown influence function. Sufficient conditions for the existence of such a representation are provided. Similarly, the nonlinear function estimate is generated by an integral of the predefined kernel multiplied by an influence function estimate. Using the time history of the plant, the learning rule indirectly estimates the unknown function by updating the influence function estimate. It is shown that the estimate function converges to the actual disturbance asymptotically. Consequently, the controller achieves the disturbance cancellation asymptotically. The method is extended to repetitive control applications. It is applied to the control of robot manipulators. Simulation and actual real-time implementation results using the Berkeley/NSK robot arm show that the proposed learning algorithm is more robust and converges at a faster rate than conventional repetitive controllers. >

Comparison of four discrete-time repetitive control algorithms

This paper is a much abbreviated version of an extensive comparison of four different repetitive control algorithms. Performance issues such as computational intensity and execution time are discussed. These algorithm were used on a computer disk drive. From presented experimental data conclusions are drawn about convergence and robustness.

On compensator design for linear time-invariant dual-input single-output systems

This paper presents a new method for the design of compensators for linear time-invariant dual-input/single-output (DISO) systems in continuous time or discrete time. The new method reduces the problem to two single-input/single-output (SISO) design problems, which are well suited to frequency-response design techniques. The first part of the method is the design of a stabilizing compensator for an auxiliary feedback system. The auxiliary compensator parameterizes the two output blocks of the single-input/dual-output compensator such that the zeros of the parallel system formed by cascade of the compensator with the plant are stable. The auxiliary compensator also determines the relative contribution to the output of the two parallel subsystems of the DISO system. The second SISO compensator design is used to ensure that the feedback system is stable and that performance and robustness specifications are achieved. This paper includes a discrete-time-design example for a dual-stage actuator system for a disk drive including implementation results. Straightforward extensions for multi-input/single-output systems are discussed.

A Robust Approach to High-Speed Navigation for Unrehearsed Desert Terrain

This article presents a robust approach to navigating at high-speed across desert terrain. A central theme of this approach is the combination of simple ideas and components to build a capable and robust system. A pair of robots were developed which completed a 212 kilometer Grand Challenge desert race in approximately seven hours. A path-centric navigation system uses a combination of LIDAR and RADAR based perception sensors to traverse trails and avoid obstacles at speeds up to 15m/s. The onboard navigation system leverages a human based pre-planning system to improve reliability and robustness. The robots have been extensively tested, traversing over 3500 kilometers of desert trails prior to completing the challenge. This article describes the mechanisms, algorithms and testing methods used to achieve this performance.

On controller design for linear time-invariant dual-input single-output systems

Presents a method for the design of controllers for linear time-invariant dual-input/single-output (DISO) systems in continuous-time or discrete-time. The new method reduces the problem to two single input/single-output (SISO) design problems which are well suited to frequency response design techniques. The first part of the method is the design of a stabilizing compensator for an auxiliary feedback system. The auxiliary compensator parameterizes the two output blocks of the single-input/dual-output controller such that the zeros of the parallel system formed by cascade of the controller with the plant are stable. The auxiliary compensator also determines the relative contribution to the output of the two parallel subsystems of the DISO system. The second SISO compensator design is used to ensure that the feedback system is stable and that performance and robustness specifications are achieved. The paper includes a discrete time design example for a dual-stage actuator system for a disk drive. Straightforward extensions for multi input/single-output systems are discussed.

Advanced methods for repeatable runout compensation [disc drives]

This paper presents and compares experimental results from two types of periodic disturbance compensation methods. The repeatable runout (RRO) cancellation techniques studied in this paper are adaptive feedforward cancellation (AFC) and repetitive control. Two modifications (phase advance and a feedthrough term) to the basic AFC structure are also studied experimentally. Of the AFC methods, the feedthrough technique is superior, but the repetitive controller provides better RRO rejection. Overall it is found that the removal of repeatable runout improved the tracking precision by as much as 53%. >

Cancellation of Discrete Time Unstable Zeros by Feedforward Control

This paper presents a design methodology for the cancellation of unstable zeros in linear discrete time systems with tracking control objectives. Unstable zeros are defined to be those zeros of the rational transfer function that occur outside the unit circle. Unstable zeros cannot be canceled by feedback without compromizing stability. In light of the fact, a feedforward scheme is used. Future desired trajectory information is required because the feedforward scheme is noncausal. The amount of future desired trajectory information that is required depends upon the zero locations and design specifications. It is shown that for a known plant with no zeros on the unit circle one can obtain a frequency response arbitrarily close to 1

Design and Flight Testing of an H00 Controller for a Robotic Helicopter

Although robotic helicopters have received increasing interest from university, industry, and military research groups, their flight envelope in autonomous operations remains extremely limited. The absence of high-fidelity simulation models has prevented the use of well-established multivariable control techniques for the design of high-bandwidth control systems. Existing controllers are of low bandwidth and cover only small portions of the vehicles flight envelope. The results of the synergic use of high-fidelity integrated modeling strategies and robust multivariable control techniques for the rapid and reliable design of a high-bandwidth controller for robotic helicopters are presented. The project implemented and flight tested an H ∞ loop shaping controller on the Carnegie Mellon University (CMU) Yamaha R-50 robotic helicopter. During the flight tests, the CMU R-50 flew moderate-speed coordinated maneuvers with a level of tracking performance that exceeds performance reported in the publicly available literature. The authors believe that the results open the road to the implementation on robotic helicopters of full-flight-envelope control systems for complex autonomous missions.

Exponential convergence of a learning controller for robot manipulators

The proof for the exponential convergence of a class of learning and repetitive control algorithms for robot manipulators is given. The learning process involves the identification of the robot inverse dynamics function by having the robot execute a set of tasks repeatedly. Using the concepts of functional persistence of excitation and functional uniform complete observability, it is shown that, when a training task is selected for the robot which is persistently exciting, the learning controllers are globally exponentially stable. Repetitive controllers are always exponentially stable. >

ATTITUDE CONTROL OPTIMIZATION FOR A SMALL-SCALE UNMANNED HELICOPTER

This paper presents results from the attitude control optimization for a small-scale helicopter by using an identified model of the vehicle dynamics that explicitly accounts for the coupled rotor/stabilizer/fuselage (r/s/f) dynamics. The accuracy of the model is verified by showing that it successfully predicts the performance of the control system currently used for Carnegie Mellons autonomous helicopter (baseline controller). Elementary stability analysis shows that the light damping in the coupled r/s/f mode, which is due to the stabilizer bar, limits the performance of the baseline control system. This limitation is compensated by a second order notch filter. The control system is subsequently optimized using the CONDUIT control design framework with a frequency response envelope specification, which allows the attitude control performance to be accurately specified while insuring that the lightly damped r/s/f mode is adequately compensated.