Bio-inspired robotics: efference copies and adaptive feedforward control

A multi-input multi-output adaptive feed-forward controller for vibration alleviation on a large blended wing body airliner

This paper focuses on a frequency-weighted hybrid adaptive control with application tosimultaneous precision positioning and vibration suppression of smart composite structures.Following the introduction of the structure and materials system of an active compositepanel (ACP) with a surface-mounted and two embedded piezoelectric ceramic patches, thesensor selection for the purpose of precision positioning or vibration control is discussed,and the function assignment of a laser displacement sensor as well as the embeddedpiezoelectric sensor is determined for the ACP. The frequency-weighted hybrid adaptivecontrol approach is then developed. The approach consists of an adaptive feedforwardposition controller for precision positioning, an adaptive feedforward vibrationcontroller for vibration suppression as well as an adaptive feedback controller for bothprecision positioning and vibration suppression. The frequency-weighted filtersintroduced in the approach realize the signal fusion of the two sensors. Finally,experiments are also performed for harmonic disturbances at the first two naturalfrequencies of the ACP and a random disturbance for both cases of with/without theadaptive feedforward vibration controller, demonstrating that the frequency-weightedhybrid adaptive control approach can provide satisfactory precision positioningand vibration suppression in both cases, and that better position accuracy canbe achieved if the adaptive feedforward vibration controller is also employed.

When prior knowledge of the plant controlled is available, feedforward control is preferred over feedback control for disturbance rejection.The stability characteristics of the system using feedforward control are the same as the original plant.Adaptive feedforward control design for gust loads alleviation of both a two degrees of freedom aerofoil with polynomial structural nonlinearity and a free flying flexible aircraft with large geometric wing deformations is investigated.Two dimensional thin aerofoil theory based on Theodorsen theory is used to perform aerodynamic analysis in the aerofoil aeroelastic model.The same aerodynamic model coupled with geometrically exact nonlinear beam is used to formulate the multi-disciplinary fluid-structure-flight framework for highly flexible aircraft.Gust loads analysis in terms of both discrete and continuous turbulence gusts are performed based on the aerofoil model and the free flying trim configuration of the highly flexible aircraft.Finite impulse response model is used to design an adaptive feedforward controller to alleviate the gust loads.Adaptive feedforward control shows good gust loads alleviation performance for both discrete and continuous gusts.Compared with adaptive feedback control, it exhibits a greater potential to reduce gust loads from the initial transient of the encounter.

Sinusoidal reference strategy for adaptive feedforward vibration control: numerical simulation and experimental study

A unique approach to adaptive feedforward control, namely the higher-harmonic time-averaged gradient descent algorithm (H-TAG), is presented. As opposed to the more common adaptive feedforward control method based on the filtered-x LMS algorithm, H-TAG does not require a measure of the control to error sensor transfer function for the system. Thus if the system dynamics change during the control process, H-TAG will simply adapt to the new optimal control configuration, rendering the approach model insensitive. Until now, work presented on the H-TAG algorithm has been limited to a single-input, single-output theoretical simulation. This paper describes the additional considerations involved in implementing the algorithm on an actual complex laboratory model (a de Havilland Dash 7 turbo-prop fuselage) that has multiple inputs and multiple outputs. Results from this preliminary experimental study suggest that the H-TAG algorithm is a viable alternative for adaptive feedforward control of systems with slowly time-varying parameters.

In this paper, high speed synchronization of multiple motion control axes under adaptive feedforward control is considered. The adaptive feedforward control system for one axis system consists of a proportional feedback controller and an adaptive feedforward controller. To synchronize the multiple motion control axes, the synchronizing controller, which responds to the tracking errors, is introduced. The structure of the synchronizing controller is investigated. The parameters of the adaptive feedforward controller are adjusted by the adaptation algorithm so that the tracking errors and the synchronization errors are removed. This results in the significant improvement of the tracking responses and the quick removal of the synchronization errors during transient as well as steady state. The synchronization for a four axes motion control system, of which parameters are significantly different, is investigated. Stability of the proposed synchronizing system is analyzed. Experimental results demonstrate the effectiveness of the proposed control scheme for synchronization.

In this paper we focus on the adaptive control of structural acoustics using intelligent structures with embedded piezoelectric (PZT) patches and low cost digital signal processor systems. After a discussion on the adaptive feedforward control scheme, a hybrid adaptive control scheme is proposed, which takes advantage of both feedback control and adaptive feedforward control. The two schemes are realized on a low-cost, small volume, convenient and universal digital signal processing (DSP) board. A carbon fiber reinforced polymer plate with two embedded PZT patches is developed and used in two experiments. The first experiment is adaptive interior noise control using the intelligent plate, in which the adaptive feedforward control scheme is employed. Obvious noise reduction is obtained for constant frequency, swept frequency and varying amplitude harmonic disturbances. The second experiment is adaptive control of sound-induced vibration of the plate, where two embedded PZT patches are used as an actuator and a sensor, respectively, and the hybrid adaptive controller is applied. The full vibration reduction for various harmonic excitations is obtained, verifying the advantage of the hybrid adaptive control. It is demonstrated that active control of structural acoustics can be efficiently achieved by employing intelligent structures, advanced adaptive control schemes and the low-cost DSP board.

Fatigue-resistant and graded muscle forces can be evoked through asynchronous intrafascicular multi-electrode stimulation (aIFMS). Prior studies on controlled force generation using aIFMS employed either a feedback controller featuring a multiple-input single-output delayed-integral (MISO-$\delta$I) control law, or a feedforward controller with a non-predictive model-based policy. However, these controllers resulted in lagged responses as stimulation was coordinated via intentional time delays and lacked immediate control corrections. To address these limitations, this paper presents an adaptive feedforward model predictive controller (aF-MPC) for isometric torque control. The aF-MPC was evaluated in experiments in anesthetized felines implanted with Utah Slanted Electrode Arrays in their sciatic nerves. This controller redesigned the existing aIFMS feedforward controller by enhancing it with a predictive policy and an online model learning algorithm to compensate for unaccounted aIFMS effects. Statistical comparisons of the aF-MPC and the (non-adaptive) F-MPC trials and observational comparisons of the aF-MPC and the MISO-$\delta$I controller were performed for different desired trajectories. The aF-MPC exhibited significant performance improvements over the F-MPC across multiple metrics. Observationally, the aF-MPC showed improvements in all performance metrics over the MISO-$\delta$I controller. Despite unknown dynamics in the aIFMS system, this paper's aF-MPC outperformed alternate approaches as it accurately tracked desired torque profiles even under high-frequency commands. The application of the aF-MPC in conjunction with aIFMS could provide a better avenue for developing naturalistic motor neuroprosthesis than F-MPCs or MISO-$\delta$I controllers.

Usage of frequency based error correction methods for repetitive disturbances becomes popular in data storage devices that use a rotating mechanical power plant. Two distinct disturbance cancellation methods, disturbance observer (DOB) and adaptive feed forward control (AFC) are the most popular approaches used for current drive servo integration. DOB reduces error using a predetermined function that is calculated thorough a complicated process. Generally the response to disturbance and initiation of control action of DOB is relatively fast even if the amount of error reduction is limited. Meanwhile the error elimination capability of AFC for the repetitive disturbance is normally better than DOB as it utilizes a Fourier coefficients updating feature that supports an enhanced fine adaptation of a certain frequency disturbance. Of course, AFC suffers for disturbances that have neighboring close frequencies due to the inherent fundamental frequency dependencies of AFC formulation. In the present work, an attempt for combining benefits of AFC and DOB is made. This might enables control action not only to generate fast response to disturbance but also perform an accurate error rejection. This feature also provides to the closed loop system a frequency selection for enabling a strategic selective error rejection. The proposed method named Hybrid Disturbance Observer (HDOB) is compared to the traditional AFC and DOB.

An Adaptive Feedforward Control Algorithm for Computer Control of Wastewater Neutralization

Presents an approach to the design and real-time implementation of an adaptive controller for a robotic manipulator based on digital signal processors. The Texas Instruments DSP (TMS320C31) chips are used in implementing real-time adaptive control algorithms to provide enhanced motion control performance for robotic manipulators. In the proposed scheme, adaptation laws are derived from the direct model reference adaptive control principle based on the improved Lyapunov second method. The proposed adaptive controller consists of an adaptive feedforward and feedback controller and PI-type time-varying auxiliary control elements. The proposed control scheme is simple in structure, fast in computation, and suitable for real-time control. Moreover, this scheme does not require any accurate dynamic modeling nor values of manipulator parameters and payload. Performance of the proposed adaptive controller is illustrated by simulation and experimental results for an industrial robot with four joints in the joint space and Cartesian space.

This paper describes sound transmission reduction control through thin plate by using piezoelectric ceramic sensors and actuators, both of which are bonded on both surfaces of the plate. 2-DOF control system is introduced to reject the effect of bending vibration of the periodic component and continuously distributed component over the broad frequency range. The adaptive feedforward control is realized by using FIR filters and LMS system identification algorithm. The filtered epsilon with delay system is applied to cope with the non-minimum phase of the identified feedforward control systems. The internal model control is introduced to the feedback control system, which are constructed basically with the direct velocity feedback (DVFB) system. The internal model control was efficient to suppress the howling instability to occur frequently in case of high feedback gain. The substantial reduction of the sound radiation from the plate in the control experiment proves the effectiveness of the 2 DOF control system.

Active engine mounts significantly contribute to ensure the comfort in vehicles with emission-reducing engine technologies, e.g., cylinder-on-demand (COD), downsizing or turbochargers. To control active engine mounts, either adaptive or non-adaptive feedforward control is commonly employed. Since both approaches have previously been treated separately, this study proposes methods to connect them in terms of multiple-input-multiple-output Newton/FxLMS adaptive filters with self-trained, grid-based look-up tables. The look-up tables are incorporated as parameter-maps or parallel-maps, respectively. By combining the two feedforward control strategies, their inherent advantages, i.e., the adaptivity of adaptive filtering and the direct impact as well as the tracking behavior of map-based feedforward control, are utilized. The proposed control structures are illustrated by simulation and experimentally demonstrated in a vehicle with a V8-COD engine. While both methods significantly reduce the convergence time of the adaptive filter, the parallel implementation additionally improves the tracking behavior during fast engine run-ups.

Unmanned Surface Vehicles (USVs) have widespread commercial, scientific and military applications. In some of these applications station keeping is essential. Since USVs have a bigger surface/ weight ration than large commercial ships, they are more sensitive to wind disturbances, especially during station keeping. The modeling, identification and control of a 4.9 meter long, 180 kilogram wave adaptive modular vehicle (WAM-V) USV are presented in this paper. The propulsion system of the USV is custom designed with two azimuthing thrusters, making the USV fully actuated. System identification tests were conducted to get the precise model of the USV. A sliding mode controller was first applied to the USV and a wind feedforward controller was designed to aid the nonlinear feedback control to mitigate wind disturbance. For the feedforward controller design, a wind model was formulated that uses the apparent wind speed and direction collected from a high accuracy anemometer. A model based adaptive mechanism was developed to tune the feedforward control gains for the best performance. Station keeping field tests were conducted at a location with an average wind speed of 9 knots. The results show that with the tuned feedforward controller, the USV can hold its position and heading better over extended periods of time than a single feedback controller.

The proposed investigation will lead to the development of adaptive truss technology for use in future aircraft and satellite design. The adaptive scheme is of feedforward type and essentially generates a secondary behavior using actuators to control host structure to meet the desired requirements. Three points are focused in this paper First, an adaptive controller is designed based on the filtered-X Least Mean Square (X-LMS) algorithm. Secondly, the control algorithm is developed under Matlab and Labview for simulation and real-time adaptive control. Forthermore, in the work, an existing truss structure with three sensors and three motor actuators has been used to examine the control system. The research results show us that the controller is feasible and suitable to die on-line control of adaptive taiss described as a slowing varying system. Introduction A structure in space may require changes to initially establish the desired configuration and characteristics by self-turning to adapt the variable environment. Adaptive trusses are those that incorporate actuators and sensors that are highly integrated into the host structures. These structures are capable of altering both their mechanical states (position or velocity) and their mechanical characteristics (stiffness or damping), according to time-varying environment. Once the precision structure is in the operational position, its static position must be precisely located. The desired static position can be attained by adjusting active members. The shape control of truss structures has been studied # Graduate Student (zlizhuo@hippo.nwpu.edu.cn) # Professor of Aero Engine Department ## Engineer, Department of Aerospace Department Copyright @ 2001 by the American Institute of the Aeronautics and Astronautics, Inc. All rights reserved. by several investigators in the decades. For example, optimization of the support truss was formulated so as to minimize die required effort to correct the shape, while improving the performance of the antenna. Minimization of the effort required for the shape correction was implemented in the form of the minimizing the maximum actuator input. Bruno and Salama ' have achieved significant improvements in static accuracy of the Precision Segmented Reflector (PSR) using only limited number of actuators. All of these investigations showed the feasibility of using active members to adaptively alter the shape of the structure and/or its behavior to other states. In this paper, a truss controller is developed for shape control by using X-LMS and Fuzzy Logic Control. Under Matlab and Labview, the computer simulation and physical experiments are used to examine the characteristics of the controller according to a truss structure model with 3 sensors, 3 actuators and 3 exciters. The primary results are supplied and will be discussed in the following sections. Adaptive Feedforward Controller The increase in computational capabilities of modern digital signal processing system has permitted the real-time implementation (at high bandwidths) of adaptive filter algorithms developed in the 1960s and 1970s. These algorithms provide a control system that can adapt to changes in the plant H (z) and provide outstanding disturbance rejection for certain classes of inputs. In the control scheme, an adaptive feedforward controller has been designed based on X-LMS algorithm. ' In the development of the LMS algorithm the output of the LMS filter is assumed to be the convolution of the input data sequence to the filter with the LMS filter coefficients. So, the control filter is an FIR filter. A block diagram of the controller is shown in Fig. 1. In the schematic diagram, the error output sensors consist of M outputs arranged in a vector, E(k), of length M as follows: (1) 1 American Institute of Aeronautics and Astronautics (c)2001 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)' Sponsoring Organization. me uuimui miei^ consist or an array ot P actuators Y(k] = H(k} * U(k] (9) (or control inputs), each of order N. These aggregate control weights are arranged in one vector: ^^ is ^ M x P matrix of me control to output impulse response functiopns. The matrix H(k) is n of the form: