Sandor M. Veres

Predictive self-tuning control by parameter bounding and worst-case design

Abstract The computation of bounds on the parameters of a plant model allows worst-case control synthesis, taking account of the uncertainty in the model. This paper introduces such a control scheme: predictive bounding control. The scheme contrasts with existing self-tuning control methods which base control synthesis on a nominal plant model. Parameter bounding also permits detection of abrupt plant changes and adaptive tracking of time-varying plant characteristics by suitable choice of bounds on plant-model output error and plant-parameter increments. Estimation and control are closely integrated and the control computation can compromise between reducing the model uncertainty and reducing predicted output error. Simulation examples show the excellent performance of predictive bounding control.

Universal adaptive control of satellite formation flying

In this paper, a new universal adaptive control scheme for satellite formation flying is developed. The underlying idea of our design is to combine the domination design and the monotone adaptive gain. This scheme is guaranteed to have the properties of position tracking and full adaptivity against all parameters. Simulation studies are given which establish that implementation of this scheme would not require unachievable actuator signals.

Frequency selective feedback for active noise control

Active noise control has moved from the laboratory to industrial applications. This article presents a frequency-selective, filter-based adaptive feedback solution in the frequency domain for periodic noise attenuation.

Autonomous vehicle control systems — a review of decision making

A systematic review is provided on artificial agent methodologies applicable to control engineering of autonomous vehicles and robots. The paper focuses on some fundamentals that make a machine autonomous: decision making that involves modelling the environment and forming data abstractions for symbolic processing and logic-based reasoning. Most relevant capabilities such as navigation, autonomous path planning, path following control, and communications, that directly affect decision making, are treated as basic skills of agents. Although many autonomous vehicles have been engineered in the past without using the agent-oriented approach, most decision making onboard of vehicles is similar to or can be classified as some kind of agent architecture, even if in a naïve form. First the ANSI standard of intelligent systems is recalled then a summary of the fundamental types of possible agent architectures for autonomous vehicles are presented, starting from reactive, through layered, to advanced architectures in terms of beliefs, goals, and intentions. The review identifies some missing links between computer science results on discrete agents and engineering results of continuous world sensing, actuation, and path planning. In this context design tools for ‘abstractions programming’ are identified as needed to fill in the gap between logic-based reasoning and sensing. Finally, research is reviewed on autonomous vehicles in water, on the ground, in the air, and in space with comments on their methods of decision making. One of the main conclusions of this review is that standardization of decision making through agent architectures is desirable for the future of intelligent vehicle developments and their legal certification.

Parameter-Bounding Algorithms for Linear Errors-in-Variables Models

Abstract The static and dynamic cases of parameter bounding for errors-in-variables models are discussed and their differences clarified. Algorithms to calculate parameter bounds for such models are presented. A search method and polytope and ellipsoid bounding are considered. Methods are given for finding orthants of the parameter space which contain no feasible parameter vectors and can be discarded, making the bounding algorithm faster. Simulation examples illustrate the problems and the algorithms.

Practical verification of decision-making in agent-based autonomous systems

We present a verification methodology for analysing the decision-making component in agent-based hybrid systems. Traditionally hybrid automata have been used to both implement and verify such systems, but hybrid automata based modelling, programming and verification techniques scale poorly as the complexity of discrete decision-making increases making them unattractive in situations where complex logical reasoning is required. In the programming of complex systems it has, therefore, become common to separate out logical decision-making into a separate, discrete, component. However, verification techniques have failed to keep pace with this development. We are exploring agent-based logical components and have developed a model checking technique for such components which can then be composed with a separate analysis of the continuous part of the hybrid system. Among other things this allows program model checkers to be used to verify the actual implementation of the decision-making in hybrid autonomous systems.

Verifying fault tolerance and self-diagnosability of an autonomous underwater vehicle

We report the results obtained during the verification of Autosub6000, an autonomous underwater vehicle used for deep oceanic exploration. Our starting point is the Simulink/Matlab engineering model of the submarine, which is discretised by a compiler into a representation suitable for model checking. We assess the ability of the vehicle to function under degraded conditions by injecting faults automatically into the discretised model. The resulting system is analysed by means of the model checker MCMAS, and conclusions are drawn on the systems ability to withstand faults and to perform self-diagnosis and recovery. We present lessons learnt from this and suggest a general method for verifying autonomous vehicles.

Model-free frequency domain iterative active sound and vibration control

In this paper, a model-free iterative feedback tuning method is presented to tune a frequency domain controller for a set of narrow-band disturbances. A single control problem with multiple tones of the disturbance is split into a set of parallel control problems. Due to the representation in the frequency domain, there are only control amplitude and phase to be tuned for each tone. Tuning is performed through a sequence of experiments which enables the calculation of a model-free gradient of a quadratic cost function. The number of experiments at each iterative step is R×S+1, where R is the number of controls signals and S is the number of output signals. Results on a laboratory active sound control system are shown. The example demonstrates the practical effectiveness of the method.

Terrain referencing for autonomous navigation of underwater vehicles

A terrain aided navigation correction system has been developed for use with the Autonomous Underwater Vehicle (AUV) Autosub 6000. This system allows drift in dead-reckoning navigation to be estimated by matching bathymetry obsevered in multibeam echosounder (MBE) data to a reference map. The reference map consists of a single line of bathymetric data, one swath width across, spanning the operational area of a survey mission. This map is collected by Autosub prior to undertaking the survey. This paper presents a discussion of the biases in typical AUV navigational sensors and their influence on navigation on a map that is created in-mission by a submersible. Based on this discussion a filter has been implemented and has been used to analyse the navigational errors accumulated during an Autosub 6000 survey mission and computational limitations for realtime application are assessed.

Structure identification of parameter-bounding models by use of noise-structure bounds

The selection of a model for a given set of records is discussed for the case when the output error is required to be always within some specified bounds. A new selection criterion is proposed: the model should be the simplest giving sufficiently unstructured output errors. The lack of structure is specified by requiring the candidate model to be capable of giving sufficiently small sample autocorrelations of the model-output errors over a range of lags. Computational algorithms to apply the criterion are presented, and illustrated by tests on AR processes.