Bias-free identification of a linear model-predictive steering controller from measured driver steering behavior

Recent developments in modeling driver steering control with preview are reviewed. While some validation with experimental data has been presented, the rigorous application of formal system identification methods has not yet been attempted. This paper describes a steering controller based on linear model-predictive control. An indirect identification method that minimizes steering angle prediction error is developed. Special attention is given to filtering the prediction error so as to avoid identification bias that arises from the closed-loop operation of the driver-vehicle system. The identification procedure is applied to data collected from 14 test drivers performing double lane change maneuvers in an instrumented vehicle. It is found that the identification procedure successfully finds parameter values for the model that give small prediction errors. The procedure is also able to distinguish between the different steering strategies adopted by the test drivers. © 2006 IEEE.

Bayesian model selection for time series using Markov chain Monte Carlo

We present a stochastic simulation technique for subset selection in time series models, based on the use of indicator variables with the Gibbs sampler within a hierarchical Bayesian framework. As an example, the method is applied to the selection of subset linear AR models, in which only significant lags are included. Joint sampling of the indicators and parameters is found to speed convergence. We discuss the possibility of model mixing where the model is not well determined by the data, and the extension of the approach to include non-linear model terms.

Surface excavation with model predictive control

This paper describes the application of variable-horizon model predictive control to trajectory generation in surface excavation. A nonlinear dynamic model of a surface mining machine digging in oil sand is developed as a test platform. This model is then stabilised with an inner-loop controller before being linearised to generate a prediction model. The linear model is used to design a predictive controller for trajectory generation. A variable horizon formulation is augmented with extra terms in the cost function to allow more control over digging, whilst still preserving the guarantee of finite-time completion. Simulations show the generation of realistic trajectories, motivating new applications of variable horizon MPC for autonomy that go beyond the realm of vehicle path planning. ©2010 IEEE.

Kernelized log linear models for continuous speech recognition

Large margin criteria and discriminative models are two effective improvements for HMM-based speech recognition. This paper proposed a large margin trained log linear model with kernels for CSR. To avoid explicitly computing in the high dimensional feature space and to achieve the nonlinear decision boundaries, a kernel based training and decoding framework is proposed in this work. To make the system robust to noise a kernel adaptation scheme is also presented. Previous work in this area is extended in two directions. First, most kernels for CSR focus on measuring the similarity between two observation sequences. The proposed joint kernels defined a similarity between two observation-label sequence pairs on the sentence level. Second, this paper addresses how to efficiently employ kernels in large margin training and decoding with lattices. To the best of our knowledge, this is the first attempt at using large margin kernel-based log linear models for CSR. The model is evaluated on a noise corrupted continuous digit task: AURORA 2.0. © 2013 IEEE.

Optimal control predicts human performance on objects with internal degrees of freedom.

On a daily basis, humans interact with a vast range of objects and tools. A class of tasks, which can pose a serious challenge to our motor skills, are those that involve manipulating objects with internal degrees of freedom, such as when folding laundry or using a lasso. Here, we use the framework of optimal feedback control to make predictions of how humans should interact with such objects. We confirm the predictions experimentally in a two-dimensional object manipulation task, in which subjects learned to control six different objects with complex dynamics. We show that the non-intuitive behavior observed when controlling objects with internal degrees of freedom can be accounted for by a simple cost function representing a trade-off between effort and accuracy. In addition to using a simple linear, point-mass optimal control model, we also used an optimal control model, which considers the non-linear dynamics of the human arm. We find that the more realistic optimal control model captures aspects of the data that cannot be accounted for by the linear model or other previous theories of motor control. The results suggest that our everyday interactions with objects can be understood by optimality principles and advocate the use of more realistic optimal control models for the study of human motor neuroscience.

Transient growth and triggering in the horizontal Rijke tube Matthew

This theoretical paper examines a non-normal and non-linear model of a horizontal Rijke tube. Linear and non-linear optimal initial states, which maximize acoustic energy growth over a given time from a given energy, are calculated. It is found that non-linearity and non-normality both contribute to transient growth and that, for this model, linear optimal states are only a good predictor of non-linear optimal states for low initial energies. Two types of non-linear optimal initial state are found. The first has strong energy growth during the first period of the fundamental mode but loses energy thereafter. The second has weaker energy growth during the first period but retains high energy for longer. The second type causes triggering to self-sustained oscillations from lower energy than the first and has higher energy in the fundamental mode. This suggests, for instance, that low frequency noise will be more effective at causing triggering than high frequency noise.

Investigation of self excited and acoustically forced combustion instabilities with laser diagnostics

Combustion oscillations in gas turbines can result in serious damage. One method used to predict such oscillations is to analyze the combustor acoustics using a simple linear model. Such a model requires a flame transfer function to describe the response of the heat release to flow perturbations inside the combustor. This paper reports on the application of Planar Laser Induced Fluorescence (PLIF) of OH radicals to analyze the response of a lean premixed flame to oncoming flow perturbations. Both self-excited oscillations and low amplitude forced oscillations at various frequencies are investigated in an atmospheric pressure model combustor rig. In order to visualize fluctuations of local fuel distribution, acetone-PLIF was also applied in non-reacting and acoustically forced flows at oscillation frequencies of 200 Hz and 510 Hz, respectively. OH-PLIF images were acquired over a range of operating parameters. The results presented in this paper originate from data sets acquired at fixed phase angles during the oscillation cycle. Comparative experiments in self excited and forced acoustic oscillations show that the flame and the combustion intensity develop similarly throughout the pressure cycle in both cases. Although the peak fluorescence intensities differ between self excited and the forced instabilities, there is a clear correspondence in the observed frequency and phase information from the two cases. This result encourages a comparison of the OH-PLIF and the acetone-PLIF results. Quantitative measurements of the equivalence ratio in specific areas of the measurement plane offer insight on the complex phenomena coupling acoustic perturbations, i.e. flow velocity fluctuations, to fluctuations in fuel distribution and combustion intensity, ultimately resulting in self excited combustion oscillations.

Extending noise robust structured support vector machines to larger vocabulary tasks

This paper describes a structured SVM framework suitable for noise-robust medium/large vocabulary speech recognition. Several theoretical and practical extensions to previous work on small vocabulary tasks are detailed. The joint feature space based on word models is extended to allow context-dependent triphone models to be used. By interpreting the structured SVM as a large margin log-linear model, illustrates that there is an implicit assumption that the prior of the discriminative parameter is a zero mean Gaussian. However, depending on the definition of likelihood feature space, a non-zero prior may be more appropriate. A general Gaussian prior is incorporated into the large margin training criterion in a form that allows the cutting plan algorithm to be directly applied. To further speed up the training process, 1-slack algorithm, caching competing hypothesis and parallelization strategies are also proposed. The performance of structured SVMs is evaluated on noise corrupted medium vocabulary speech recognition task: AURORA 4. © 2011 IEEE.

Vibration of a high-altitude tethered balloon with application to climate engineering

A balloon tethered at an altitude of 20 km could deliver a particulate cloud leading to global cooling. Tethering a balloon at this altitude poses significant problems with respect to vibration and stability, especially in regions of high wind. No-one has ever proposed, yet alone launched, a balloon at an altitude of 20 km tethered to the ground. Owing to wind, the tether needs to be 23 km in length and is to be fixed to a ship at sea or on land in equatorial regions. Whilst the balloon at 20 km is subject to relatively modest wind conditions, at jet stream altitudes (10km) the tether will experience much higher wind loadings, not only because of the high wind speeds of up to 300 km / hr but also because of the high air density. A tether of circular cross section in these high winds will be subject to horizontal and downward drag forces that would bring the aerostat down. For this reason it is advantageous to consider a self-aligning tether of an aerodynamic cross section whereby it is possible to reduce the drag substantially. One disadvantage of a non-circular tether is the possibility of flutter and galloping instabilities. It is reasonably straightforward to model these phenomena for short lengths of aerofoil, but the situation becomes more complex for a 20 km tensioned tether with large deflection and curvature, variable wind speed, variable air density and variable tension. Analysis using models of infinite length are used to establish the stability at a local scale where the tension, aerodynamic and geometric properties are considered constant. Dispersion curve analysis is useful here. But for dynamics on a long-wavelength scale (several km) then a full non-linear analysis is required. This non-linear model can be used to establish the local values of tension appropriate for the dispersion analysis. This keynote presentation will give some insight into these issues.

Efficient decoding with generative score-spaces using the expectation semiring

State-of-the-art speech recognisers are usually based on hidden Markov models (HMMs). They model a hidden symbol sequence with a Markov process, with the observations independent given that sequence. These assumptions yield efficient algorithms, but limit the power of the model. An alternative model that allows a wide range of features, including word- and phone-level features, is a log-linear model. To handle, for example, word-level variable-length features, the original feature vectors must be segmented into words. Thus, decoding must find the optimal combination of segmentation of the utterance into words and word sequence. Features must therefore be extracted for each possible segment of audio. For many types of features, this becomes slow. In this paper, long-span features are derived from the likelihoods of word HMMs. Derivatives of the log-likelihoods, which break the Markov assumption, are appended. Previously, decoding with this model took cubic time in the length of the sequence, and longer for higher-order derivatives. This paper shows how to decode in quadratic time. © 2013 IEEE.

Reconfigurable predictive control for redundantly actuated systems with parameterised input constraints

A method is proposed for on-line reconfiguration of the terminal constraint used to provide theoretical nominal stability guarantees in linear model predictive control (MPC). By parameterising the terminal constraint, its complete reconstruction is avoided when input constraints are modified to accommodate faults. To enlarge the region of feasibility of the terminal control law for a certain class of input faults with redundantly actuated plants, the linear terminal controller is defined in terms of virtual commands. A suitable terminal cost weighting for the reconfigurable MPC is obtained by means of an upper bound on the cost for all feasible realisations of the virtual commands from the terminal controller. Conditions are proposed that guarantee feasibility recovery for a defined subset of faults. The proposed method is demonstrated by means of a numerical example. © 2013 Elsevier B.V. All rights reserved.