Kevin M. Passino

Biomimicry of bacterial foraging for distributed optimization and control

We explain the biology and physics underlying the chemotactic (foraging) behavior of E. coli bacteria. We explain a variety of bacterial swarming and social foraging behaviors and discuss the control system on the E. coli that dictates how foraging should proceed. Next, a computer program that emulates the distributed optimization process represented by the activity of social bacterial foraging is presented. To illustrate its operation, we apply it to a simple multiple-extremum function minimization problem and briefly discuss its relationship to some existing optimization algorithms. The article closes with a brief discussion on the potential uses of biomimicry of social foraging to develop adaptive controllers and cooperative control strategies for autonomous vehicles. For this, we provide some basic ideas and invite the reader to explore the concepts further.

Decentralized adaptive control of nonlinear systems using radial basis neural networks

Stable direct and indirect decentralized adaptive radial basis neural network controllers are presented for a class of interconnected nonlinear systems. The feedback and adaptation mechanisms for each subsystem depend only upon local measurements to provide asymptotic tracking of a reference trajectory. Due to the functional approximation capabilities of radial basis neural networks, the dynamics for each subsystem are not required to be linear in a set of unknown coefficients as is typically required in decentralized adaptive schemes. In addition, each subsystem is able to adaptively compensate for disturbances and interconnections with unknown bounds.

Biomimicry of Social Foraging Bacteria for Distributed Optimization: Models, Principles, and Emergent Behaviors

In this paper, we explain the social foraging behavior of E. coli and M. xanthus bacteria and develop simulation models based on the principles of foraging theory that view foraging as optimization. This provides us with novel models of their foraging behavior and with new methods for distributed nongradient optimization. Moreover, we show that the models of both species of bacteria exhibit the property identified by Grunbaum that postulates that their foraging is social in order to be able to climb noisy gradients in nutrients. This provides a connection between evolutionary forces in social foraging and distributed nongradient optimization algorithm design for global optimization over noisy surfaces.

Fuzzy model reference learning control for cargo ship steering

The use of a learning control system to maintain adequate performance of a cargo ship autopilot when there are process disturbances or variations is examined. The objective is to make an initial assessment of what advantages a fuzzy learning control approach has over conventional adaptive control approaches. The simulation results indicate that the fuzzy model reference learning controller (FMRLC) has several potential advantages over model reference adaptive control (MRAC), including improved convergence rates, use of less control energy, enhanced disturbance rejection properties, and lack of dependence on a mathematical model. Using the comparative analysis, the authors discuss how the well-developed concepts in conventional adaptive control can be used to evaluate fuzzy learning control techniques. >

Fuzzy learning control for antiskid braking systems

Although antiskid braking systems (ABS) are designed to optimize braking effectiveness while maintaining steerability, their performance often degrades under harsh road conditions (e.g. icy/snowy roads). The use of the fuzzy model reference learning control (FMRLC) technique for maintaining adequate performance even under such adverse road conditions is proposed. This controller utilizes a learning mechanism that observes the plant outputs and adjusts the rules in a direct fuzzy controller so that the overall system behaves like a reference model characterizing the desired behavior. The performance of the FMRLC-based ABS is demonstrated by simulation for various road conditions (wet asphalt, icy) and transitions between such conditions (e.g. when emergency braking occurs and the road switches from wet to icy or vice versa). >

Multiple task assignments for cooperating uninhabited aerial vehicles using genetic algorithms

A problem of assigning cooperating uninhabited aerial vehicles to perform multiple tasks on multiple targets is posed as a new combinatorial optimization problem. A genetic algorithm for solving such a problem is proposed. The algorithm allows us to efficiently solve this NP-hard problem that has prohibitive computational complexity for classical combinatorial optimization methods. It also allows us to take into account the unique requirements of the scenario such as task precedence and coordination, timing constraints, and trajectory limitations. A matrix representation of the genetic algorithm chromosomes simplifies the encoding process and the application of the genetic operators. The performance of the algorithm is compared to that of deterministic branch and bound search and stochastic random search methods. Monte Carlo simulations demonstrate the viability of the genetic algorithm by showing that it consistently and quickly provides good feasible solutions. This makes the real time implementation for high-dimensional problems feasible.

Towards intelligent autonomous control systems: Architecture and fundamental issues

Autonomous control systems are designed to perform well under significant uncertainties in the system and environment for extended periods of time, and they must be able to compensate for system failures without external intervention. Intelligent autonomous control systems use techniques from the field of artificial intelligence to achieve this autonomy. Such control systems evolve from conventional control systems by adding intelligent components, and their development requires interdisciplinary research. A hierarchical functional intelligent autonomous control architecture is introduced here and its functions are described in detail. The fundamental issues in autonomous control system modelling and analysis are discussed.

Fuzzy learning control for a flexible-link robot

There are two main drawbacks in fuzzy control: 1) the design of fuzzy controllers is usually performed in an ad hoc manner where it is often difficult to choose some of the controller parameters; and 2) the fuzzy controller constructed for the nominal plant may later perform inadequately if significant and unpredictable plant parameter variations occur. In this paper we illustrate these two problems on a two-link flexible robot testbed by: 1) developing, implementing, and evaluating a fuzzy controller for the robotic mechanism, and 2) illustrating that payload variations can have negative effects on the performance of a well designed fuzzy control system. Next, we show how to develop and implement a fuzzy model reference learning controller for the flexible robot and illustrate that it can automatically synthesize a rule-base for a fuzzy controller that will achieve comparable performance to the case where it was manually constructed, and automatically tune the fuzzy controller so that it can adapt to variations in the payload. >

Adaptive control of a class of decentralized nonlinear systems

Within this brief paper, a stable indirect adaptive controller is presented for a class of interconnected nonlinear systems. The feedback and adaptation mechanisms for each subsystem depend only upon local measurements to provide asymptotic tracking of a reference trajectory. In addition, each subsystem is able to adaptively compensate for disturbances and interconnections with unknown bounds. The adaptive scheme is illustrated through the longitudinal control of a string of vehicles within an automated highway system (AHS).

Intelligent Control for an Acrobot

The acrobot is an underactuated two-link planar robot that mimics the human acrobat who hangs from a bar and tries to swing up to a perfectly balanced upside-down position with his/her hands still on the bar. In this paper we develop intelligent controllers for swing-up and balancing of the acrobot. In particular, we first develop classical, fuzzy, and adaptive fuzzy controllers to balance the acrobot in its inverted unstable equilibrium region. Next, a proportional-derivative (PD) controller with inner-loop partial feedback linearization, a state-feedback, and a fuzzy controller are developed to swing up the acrobot from its stable equilibrium position to the inverted region, where we use a balancing controller to ‘catch’ and balance it. At the same time, we develop two genetic algorithms for tuning the balancing and swing-up controllers, and show how these can be used to help optimize the performance of the controllers. Overall, this paper provides (i) a case study of the development of a variety of intelligent controllers for a challenging application, (ii) a comparative analysis of intelligent vs. conventional control methods (including the linear quadratic regulator and feedback linearization) for this application, and (iii) a case study of the development of genetic algorithms for off-line computer-aided-design of both conventional and intelligent control systems.