Marina Tharayil

A survey of iterative learning control

This article surveyed the major results in iterative learning control (ILC) analysis and design over the past two decades. Problems in stability, performance, learning transient behavior, and robustness were discussed along with four design techniques that have emerged as among the most popular. The content of this survey was selected to provide the reader with a broad perspective of the important ideas, potential, and limitations of ILC. Indeed, the maturing field of ILC includes many results and learning algorithms beyond the scope of this survey. Though beginning its third decade of active research, the field of ILC shows no sign of slowing down.

Modeling and control for smart Mesoflap aeroelastic control

This paper introduces a novel concept termed Smart Mesoflaps for Aeroelastic Recirculation Transpiration (SMART) to render mass and momentum transfer for controlling shock/boundary-layer interactions in supersonic jet inlets. The SMART concept consists of a matrix of small flaps designed to undergo local aeroelastic deflection to achieve proper mass bleed or injection when subjected to shock loads. To optimize the performance of this system, NiTi shape memory alloy is used as an actuator for the flaps to control the amount of recirculation. The focus of this paper will be the subsystem modeling and control of a single flap. After a relatively detailed model is developed, a simpler model is generated, and it is experimentally shown that this approximation is adequate for control purposes. Next, the control strategy for this subsystem, subject to hysteresis and actuator saturation, is presented. A basic proportional integral derivative (PID) controller is enhanced using a hysteresis compensator (HC) and an error governor (EG). A generalized error governing scheme for PID controllers to compensate for actuator saturations is also developed. This EG method is generalizable to any stable process controlled by a PID. Finally, the PID with HC and the error governing method is experimentally applied to a benchtop SMART subsystem.

A generalized PID error governing scheme for SMART/SBLI control

Presents the control strategy used in a novel concept termed smart mesoflaps for aeroelastic recirculation transpiration (SMART) aimed at controlling shock/boundary-layer interactions (SBLI) in supersonic jet inlets. A basic PID controller used initially effectively meets the performance requirements. However, this sometimes results in actuation saturation, which can be very damaging to the SMART hardware. Therefore, a generalized error governing scheme for PID controllers is developed to compensate for actuator saturations. This method can be generalized to any stable process controlled by a PID, and has advantages over traditional anti-windup controllers. The PID error governing method is experimentally applied to a benchtop SMART subsystem and is shown to be successful at preventing excessive control efforts.

Optimizing Learning Convergence Speed and Converged Error for Precision Motion Control

This brief paper considers iterative learning control (ILC) for precision motion control (PMC) applications. This work develops a methodology to design a low pass filter, called the Q-filter, that is used to limit the bandwidth of the ILC to prevent the propagation of high frequencies in the learning. A time-varying bandwidth Q-filter is considered because PMC reference trajectories can exhibit rapid changes in acceleration that may require high bandwidth for short periods of time. Time-frequency analysis of the initial error signal is used to generate a shape function for the bandwidth profile. Key parameters of the bandwidth profile are numerically optimized to obtain the best tradeoff in converged error and convergence speed. Simulation and experimental results for a permanent-magnet linear motor are included. Results show that the optimal time-varying Q-filter bandwidth provides faster convergence to lower error than the optimal time-invariant bandwidth.

Semi active internal model control for passive disturbance rejection

This paper presents novel results on the asymptotic rejection of periodic disturbances utilizing no external power sources. The exposition is done for Single Input, Single Output linear time invariant systems that have a passive characteristic; representative of many vibration control schemes. The Internal Model Control approach is utilized to develop a controller that will asymptotically track reference commands or reject disturbances. The passivity of the open loop system is then utilized to determine a controller that acts in a semi-active fashion for the disturbance rejection case only. Two issues are addressed: overall closed loop stability and asymptotic rejection of the disturbance. The resulting controller leads to a discontinuous dynamic system and methods are given for handling this. Experimental results on a Single Degree of Freedom electromechanical testbed verify the performance and advantages of the Send Active Internal Model (SAIM) method. By carefully modulating the rate of energy dissipation, performance levels akin to fully active vibration control can be achieved without utilizing external power.

Design and convergence of a time-varying iterative learning control law

This paper presents a novel linear time-varying (LTV) iterative learning control law that can provide additional performance while maintaining the robustness and convergence properties comparable to those obtained using traditional frequency domain design techniques. Design aspects of causal and non-causal linear time-invariant (LTI), along with the proposed LTV, ILC update laws are discussed and demonstrated using a simplified example. Asymptotic as well as monotonic convergence, robustness and performance characteristics of such systems are considered, and an equivalent condition to the frequency domain convergence condition is presented for the time-varying ILC. Lastly the ILC algorithm developed here is implemented on a Microscale Robotic Deposition system to provide experimental verification.Copyright

An agent-based model for crowdsourcing systems

Crowdsourcing is a complex system composed of many interactive distributed agents whom we have little information about. Agent-based modeling (ABM) is a natural way to study complex systems since they share common properties, such as the global behavior emerging on the basis of local interactions between elements. Although significant attention has been given to dynamics of crowdsourcing systems, relatively little is known about how workers react to varying configurations of tasks. In addition, existing ABMs for crowdsourcing systems are theoretical, and not based on data from real crowdsourcing platforms. The focus of this paper is on capturing the relationships among properties of tasks, characteristics of workers, and performance metrics via an ABM. This approach is validated by running experiments on Amazon Mechanical Turk (AMT).

A Time-Varying Q-Filter Design for Iterative Learning Control

Linear time-invariant (LTI) lowpass Q-filters are often employed in iterative learning control (ILC) algorithms to provide robustness to model uncertainty at the expense of learning bandwidth. Whenever high frequency control is necessary, such as motion commands with rapid changes, precision tracking requirements may not be met. In this work, we examine the use of linear time-varying (LTV) Q-filters to create a time-varying learning bandwidth. Single-input single- output LTI plants with repeating disturbances are considered. Stability analysis for the LTV Q-filter learning algorithm is developed and an LTV Q-filter design procedure is presented incorporating time-frequency analysis of the tracking error and optimization. This procedure is used to design an LTV Q- filter for a microscale robotic deposition manufacturing system. Simulation and experimental results are provided, which demonstrate that the designed LTV Q-filter results in faster convergence and lower converged error than the best LTI Q-filter.

ACTIVE SUPERSONIC FLOW CONTROL USING HYSTERESIS COMPENSATION AND ERROR GOVERNOR

Abstract This paper introduces a novel concept termed Smart Mesoflaps for Aeroelastic Recirculation Transpiration (SMART) for controlling shock/boundary-layer interactions (SBLI) in supersonic jet inlets. The control strategy for a subsystem of the SMART project, subject to hysteresis and actuator saturation, is presented. A Hysteresis Compensation scheme, as well as results from experimental application of HC to the SMART system are presented next. A generalized error governing scheme for PID controllers to compensate for actuator saturations is also developed. Finally, the PID with HC and the error governing method is experimentally applied to a benchtop SMART subsystem and is shown to be successful at preventing excessive control efforts.

3DOF closed loop sheet alignment on non-holonomic printer differential drive registration device using input-state linearization

Aligning sheets correctly before imaging is a critical function in high-end cut sheet printers. This paper describes a differential drive electronic cut sheet registration system used in the Xerox iGen3 digital press. The sheet registration problem and system model are explained. An input-state linearization based nonlinear controller is designed to control this underactuated and non-holonomic system. Simulation and experimental results show the applicability of this controller.