Brains display very high-level parallel computation, fault-tolerance, and adaptability, all of which are what we struggle to recreate in engineered systems. The neurocomputer (an organic computer built from living neurons) seems possible and may lead to a new generation of computing device that can operate in a brain-like manner. Cultured neuronal networks on multi-electrode arrays (MEAs) are one of the best candidates for the neurocomputer for their controllability, accessibility, flexibility, and the ability to self-organize. I explored the possibility of the neurocomputer by studying whether we can show goal-directed learning, one of the most fascinating behavior of brains, in cultured networks. Inspired by the brain, which needs to be embodied in some way and interact with its surroundings in order to give a purpose to its activities, we have developed tools for closing the sensory-motor loop between a cultured network and a robot or an artificial animal (an animat), termed a ¡§hybrot¡¨. In order to efficiently find an effective closed-loop design among infinite potential options, I constructed a biologically-inspired simulated network. By using this simulated network, I designed: (1) a statistic that can effectively and efficiently decode network functional plasticity, and (2) feedback stimulations and an adaptive training algorithm to encode sensory information and to direct network plasticity. By closing the sensory-motor loop with these decoding and encoding designs, we successfully demonstrated a simple adaptive goal-directed behavior: learning to move in a user-defined direction, and further showed that multiple tasks could be learned simultaneously. These results suggest that even though a cultured network lacks the 3-D structure of the brain, it still can be functionally shaped and show meaningful behavior. To our knowledge, this is the first demonstration of goal-directed learning in embodied cultured networks. Extending from these findings, I further proposed a research plan to optimize closed-loop designs for evaluating the maximal learning capacity (or even true intelligence) of the cultured network. Knowledge gained from effective closed-loop designs provides insights about learning and memory in the nervous system, which could influence the design of neurocomputers, future artificial neural networks, and more effective neuroprosthetics.

In this thesis an adaptive backstepping control scheme is proposed for a class of multi-input perturbed systems with time-varying delays to solve regulation problems. The systems to be controlled contain n blocks¡¦ dynamic equations, hence n-1 virtual input controllers are designed from the first block to the (n-1)th block, and the backstepping controller is designed from the last block. In addition, adaptive mechanisms are embedded in each virtual input controllers and proposed controller, so that the least upper bounds of perturbations are not required to be known beforehand. Furthermore, the dynamic equations of the systems to be controlled need not satisfy strict-feedback form, and the upper bounds of the time delays as well as their derivatives need not to be known in advance either. The resultant controlled systems guarantee asymptotic stability in accordance with the Lyapunov stability theorem. Finally, a numerical example and a practical application are given for demonstrating the feasibility of the proposed control scheme.

It is well known that the Preisach model is useful to approximate the effect of hysteresis behavior in smart materials, such as piezoactuators and Shape Memory Alloy(SMA) wire actuators. For tracking control, many researchers estimate a Preisach model and then compute its inverse model for hysteresis compensation. However, the inverse of its hysteresis behavior also shows hysteresis behavior. From this idea, the inverse model with Kransnoselskii-Pokrovskii(KP) model, a developed version of Preisach model, can be used directly for SMA position control and avoid the inverse operation. Also, we propose another method for the tracking control by approximating the inverse model using an orthogonal polynomial network. To estimate and update the weight parameters in both inverse models, a gradient-based learning algorithm is used. Finally, for the SMA position control, PID controller, adaptive controllers with KP model and adaptive nonlinear inverse model controller are compared experimentally.

A simple design methodology of optimal model reference adaptive control (OMRAC) scheme with perturbation estimation for solving robust tracking problems is proposed in this thesis. The plant to be controlled belongs to a class of MIMO perturbed dynamic systems with input nonlinearity and time varying delay. The proposed robust tracking controller with a perturbation estimation scheme embedded is designed by using Lyapunov stability theorem. The control scheme contains three types of controllers. The first one is a linear feedback optimal controller, which is designed under the condition that no perturbation exists. The second one is an adaptive controller, it is used for adapting the unknown upper bound of perturbation estimation error. The third one is the perturbation estimation mechanism. The property of uniformly ultimately boundness is proved under the proposed control scheme, and the effects of each design parameter on the dynamic performance is also analyzed. An example is demonstrated for showing the feasibility of the proposed control scheme.

The dissertation addresses direct adaptive control frameworks for Lyapunov stabilization of the MIMO nonlinear uncertain systems for both uncertain discrete-time and continuous-time systems. For system theory, the development of continuous-time theory always comes along with its discrete-time counterpart. However, for direct adaptive control frameworks we find relative few Lyapunov-based results published, which is mainly due to difficulty to find feasible Lyapunov candidates and to prove negative definiteness of the Lyapunov difference. Furthermore, digital computer is widely used in all fields. Most of time, we have to deal with the direct source of discrete-time signals, even the discrete-time signals arise from continuous-time settings as results of measurement or data collection process. These motivate our study in this field. For discrete-time systems, we have investigated the results with trajectory dependent hypothesis, where the Lyapunov candidate function V combines the information from the current state k and one step ahead k-1 along the track x(k), for k≥0. The proposed frameworks guarantee partial stability of the closed-loop systems, such that the feedback gains stabilize the closed-loop system without the knowledge of the system parameters. In addition, our results show that the adaptive feedback laws can be characterized by Kronecker calculus. Later, we release this trajectory dependent hypothesis for normal discrete-time nonlinear systems. At the same time, the continuous-time cases are also studied when system with matched disturbances, where the disturbances can be characterized by known continuous function matrix and unknown parameters. Here, the trajectory dependent Lyapunov candidates (tdLC), so long as the time step |t(k)-t(k-1) | ≤ £_ and the corresponding track |x(k)-x(k-1)| ≤ £` are sufficiently small, only exist in discrete-time case. In addition, we have extended the above control designs to systems with exogenous disturbances and £d2 disturbances. Finally, we develop a robust direct adaptive control framework for linear uncertain MIMO systems under the variance of unknow system matrix from given stable solution is bounded, that is |A-Ac| ¡Ý |B Kg| ≤ |£GA|. In general, through Lyapunov-based design we can obtain the global solutions and direct adaptive control design can simultaneously achieve parameter estimation and closed-loop stability.

This thesis concerns the design, analysis and testing of adaptive controllers for rotary wing aircraft, in particular helicopters. A non-linear helicopter model is developed and validated by trim and dynamic response analyses. A inner-outer loop cascade controller is designed with a trajectory generator in the most outer layer and an adaptive neural network controller is implemented to the inner loop. Controller is then challenged to carry out complex maneuvers autonomously under turbulence. Finally, the center of gravity location is varied to severe values to observe adaptation characteristics to investigate the requirement on the knowledge of the center of gravity location during such adaptive controller design.

Differentiated-Service networks is designed for solving scalability problems through traffic aggregation. However, it can't guarantee end-to-end QoS of individual flow. In this thesis, we propose a Self-Adaptive Control Scheme for Differentiated-Service networks that can improve the throughput of individual flows dynamically. In this scheme, egress routers monitor the average throughput of individual flow, and send the Self-Adaptive Control Messages to ingress routers if need. The ingress router re-allocate network resources to improve throughput of high-priority flows depending on the Control Messages. We use NS-2 simulator to prove that our scheme that can improve throughput of high-priority flows dynamically, and suggest that a better time interval of Self-Adaptive control can be determined based on the queue sizes, packets arrival rate and departure rate. Finally, we use Random Early Detection (RED) queue instead of Drop-Tail queue to reduce unfairness of individual flows when there are congestion and insufficient network resources.

This thesis describes the implementation of Yao’s adaptive robust control to an aircraft control system. This control law is implemented as a means to maintain stability and tracking performance of the aircraft in the face of failures and changing aerodynamic response. The control methodology is implemented as an outer loop controller to an aircraft under nonlinear dynamic inversion control. The adaptive robust control methodology combines the robustness of sliding mode control to all types of uncertainty with the ability of adaptive control to remove steady state errors. A performance measure is developed in to reflect more subjective qualities a pilot would look for while flying an aircraft. Using this measure, comparisons of the adaptive robust control technique with the sliding mode and adaptive control methodologies are made for various failure conditions. Each control methodology is implemented on a full envelope, high fidelity simulation of the F-15 IFCS aircraft as well as on a lower fidelity full envelope F-5A simulation. Adaptive robust control is found to exhibit the best performance in terms of the introduced measure for several different failure types and amplitudes.

Based on the Lyapunov stability theorem, a methodology of designing an nth order adaptive integral variable structure derivative estimator (AIVSDE) is proposed in this thesis. The proposed derivative estimator not only is an improved version of the existing AIVSDE, but also can be used to estimate the nth derivative of a smooth signal which has continuous and bounded derivatives up to n+1. Analysis results show that adjusting some of the parameters can facilitate the derivative estimation of signals with higher frequency noise. The adaptive algorithm is incorporated in the estimation scheme for tracking the unknown upper bounded of the input signal as well as their's derivatives. The stability of the proposed derivative estimator is guaranteed, and the comparison between recently proposed derivative estimator of high-order sliding mode control and AIVSDE is also demonstrated.

Thesis--Auburn University, 2009. / Abstract. Vita. Includes MatLab code. Includes bibliography (p.43-44).