John H. Lilly

Adaptive tracking for pneumatic muscle actuators in bicep and tricep configurations

Adaptive tracking techniques are applied to pneumatic muscle actuators arranged in bicep and tricep configurations. The control objective is to force the joint angle to track a specified reference path. Mathematical models are derived for the bicep and tricep configurations. The models are nonlinear and in general time-varying, making adaptive control desirable. Stability results are derived, and the results of simulation studies are presented, contrasting the nonlinear adaptive control to a nonadaptive PID control approach.

Evolutionary design of a fuzzy classifier from data

Genetic algorithms show powerful capabilities for automatically designing fuzzy systems from data, but many proposed methods must be subjected to some minimal structure assumptions, such as rule base size. In this paper, we also address the design of fuzzy systems from data. A new evolutionary approach is proposed for deriving a compact fuzzy classification system directly from data without any a priori knowledge or assumptions on the distribution of the data. At the beginning of the algorithm, the fuzzy classifier is empty with no rules in the rule base and no membership functions assigned to fuzzy variables. Then, rules and membership functions are automatically created and optimized in an evolutionary process. To accomplish this, parameters of the variable input spread inference training (VISIT) algorithm are used to code fuzzy systems on the training data set. Therefore, we can derive each individual fuzzy system via the VISIT algorithm, and then search the best one via genetic operations. To evaluate the fuzzy classifier, a fuzzy expert system acts as the fitness function. This fuzzy expert system can effectively evaluate the accuracy and compactness at the same time. In the application section, we consider four benchmark classification problems: the iris data, wine data, Wisconsin breast cancer data, and Pima Indian diabetes data. Comparisons of our method with others in the literature show the effectiveness of the proposed method.

Sliding mode tracking for pneumatic muscle actuators in opposing pair configuration

Sliding mode techniques are applied to pneumatic muscle actuators arranged in an agonist/antagonist, or opposing pair configuration. The pneumatic muscle (PM) pair actuates a planar elbow manipulator, with PMs in place of bicep and tricep. The control objective is elbow angle tracking under load. A nonlinear mathematical model is derived for this system and a sliding mode controller is designed to give elbow angle tracking to within a guaranteed accuracy despite modeling errors. Static pressure requirements are also derived for stable arm behavior in the absence of a control signal. Stability results are derived, and the results of simulation studies are presented. The simulation studies also address the effects of PM heating.

Fuzzy PD+I learning control for a pneumatic muscle

A fuzzy learning control technique is used for position tracking involving the vertical movement of a mass attached to a pneumatic muscle. Because the pneumatic muscle is nonlinear and time varying, conventional fixed controllers are less effective than the fuzzy controller proposed in this paper. The controller is of a PID type, with an adaptive fuzzy PD part and a nonfuzzy integral branch. A novelty of the controller is that the fuzzy inverse model, which dynamically adjusts the PD part of the controller, incorporates the internal PM pressure as an input. Experimental results are presented from the pneumatic muscle test facility in the Human Effectiveness Laboratory at Wright-Patterson Air Force Base.

A two-input sliding-mode controller for a planar arm actuated by four pneumatic muscle groups

Multiple-input sliding-mode techniques are applied to a planar arm actuated by four groups of pneumatic muscle (PM) actuators in opposing pair configuration. The control objective is end-effector tracking of a desired path in Cartesian space. The inputs to the system are commanded input pressure differentials for the two opposing PM groups. An existing model for the muscle is incorporated into the arm equations of motion to arrive at a two-input, two-output nonlinear model of the planar arm that is affine in the input and, therefore, suitable for sliding-mode techniques. Relationships between static input pressures are derived for suitable arm behavior in the absence of a control signal. Simulation studies are reported.

Evolution of a Negative-Rule Fuzzy Obstacle Avoidance Controller for an Autonomous Vehicle

A fuzzy obstacle avoidance controller is designed for an autonomous vehicle. The controller is given the capability for obstacle avoidance by using negative fuzzy rules in conjunction with traditional positive ones. Negative fuzzy rules prescribe actions to be avoided rather than performed. A rule base of positive rules is specified by an expert for directing the vehicle to the target in the absence of obstacles, while a rule base of negative rules is experimentally determined from expert operation of the vehicle in the presence of obstacles. The consequents of the negative-rule system are codified into a chromosome, and this chromosome is evolved using an evolutionary algorithm. The resulting fuzzy system has far fewer rules than would be necessary for an obstacle avoidance controller using purely positive rules, while in addition retaining greater interpretability.

Sliding mode tracking for pneumatic muscle actuators in bicep/tricep pair configuration

This paper presents a robust motion control of pneumatic muscle actuators arranged in bicep/tricep pair configuration to deal with system and environmental uncertainties using the sliding mode approach. A mathematical model is derived for an arm with PMs in bicep/tricep pair configuration. A sliding mode controller is designed to yield asymptotic tracking of elbow angle. The results of simulations are presented to demonstrate the success of the proposed controller.

Inverse control of nonlinear systems using neural network observer and inverse mapping approach

This paper introduces a new approach to inverse control. Unlike using commonly known method of plant inverse dynamics learning, the control sequence is calculated using inverse mapping approach. Both methods are compared with nonlinear plant examples and selected desired waveforms.

Pruning via Dynamic Adaptation of the Forgetting Rate in Structural Learning

Structural learning adds a constant weight decay term to the standard backpropagation update rule. At the conclusion of training an initially oversized multilayer feedforward neural network has typically been reduced to near the minimum size required to accomplish a desired mapping and thus this method offers effective pruning. The degree of weight decay is determined by a forgetting rate E which is critical to both learning and pruning success. The choice of E requires careful consideration and current selection criteria are either inexact or computationally expensive. This paper considers E to be a dynamic parameter which can be used to control learning in order to provide redundant weights sufficient time to decay to near zero values. This approach yields pruning results comparable or superior to those obtained by considering E to be constant while reducing computational expense and eliminating the need to determine an optimal value of E .

Incorporation, characterization, and conversion of negative rules into fuzzy inference systems

This paper considers the incorporation of negative examples into fuzzy inference systems (FIS). A new method of defuzzification called dot attenuation is presented. This is a generalization of conventional defuzzification that has the ability to incorporate negative examples into the FIS reasoning process. Several variations of dot attenuation including dot product attenuation (DPA), dot minimum attenuation, and dot difference attenuation (DDA), are presented and incorporated into the center of gravity and center average defuzzification. DPA is illustrated with an inverted pendulum controller, which has a negative rule added to its rule base. The modification of the control surface due to the introduction of the negative rule is investigated. Simple steering control of a robot in the presence of obstructions using DDA is demonstrated. A method of conversion from a mixed positive/negative rule base into a standard rule base using modus tollens is introduced. Expert and automated creation of negative rules is discussed.