Alistair Armitage

An investigation into the improvement of local minima of the Hopfield network

The paper investigates the improvement of local minima of the Hopfield network. A local minima escape algorithm (LME algorithm), is proposed for improving local minima of small-scale networks. Experiments on travelling salesman problems (TSP) show that the LME algorithm is an efficient algorithm in improving the local minima, and the comparison with the simulated annealing algorithm (SA) shows that the LME algorithm can produce better results in less time. The paper then investigates the improvement of local minima of large-scale networks. By combining the LME algorithm with a network partitioning technique, a network partitioning algorithm (NPA) is proposed. Experiments on 51 and 101-city TSP problems show that the local minima of large-scale networks can be greatly improved by the NPA algorithm, however, the global minima are still difficult to achieve. Copyright 1996 Elsevier Science Ltd

USING LOW-COST INFRARED DETECTORS TO MONITOR MOVEMENT OF PEDESTRIANS: INITIAL FINDINGS

Monitoring the movement of pedestrians in everyday environments is difficult, especially if accurate data are required. Typically, data pertaining to the origins and destinations of pedestrians as they move around a space can be obtained only by either locating many observers in the area that is under surveillance or analyzing closed-circuit television footage. The former is error prone and perturbs the space being analyzed; the latter is expensive in regard to the technology and time required. Technical advances in low-cost infrared detectors provide an opportunity to observe pedestrian spaces unobtrusively and determine individual pedestrian trajectories automatically. The underpinning detector technology is described, and the way arrays of such detectors can be used to monitor larger spaces is shown. An outline of the algorithm used to create complete trajectories as pedestrians move between detectors is presented. A series of experiments is described: pedestrians were asked to move in a set of defined patterns in a controlled environment. Initial results from these experiments are discussed. It was found that at Fruin Levels of Service A to C, 93% of pedestrian trajectories could be tracked; that count dropped to 79% at Level of Service D.

A comparative study of integrated ferroelectric polymer pyroelectric sensors

Abstract This paper presents a comparative study of integrated pyroelectric sensors based on ferroelectric polymers. Two deposition technologies of PVDF and VDF/TrFE copolymer, both compatible with IC manufacturing, are presented. A 9 μm PVDF film is bonded on a silicon substrate using a capacitive acrylic thin film and a i μm VDF/TrFE copolymer is coated directly using a spin-coating technique. The theoretical response of both PVDF and VDF/TrFE pyroelectric sensors are calculated and related to the measured values. Peak responsivities of 4.5 V/W at 200 Hz and 3 V/W at 1kHz have been achieved for the pyroelectric sensors based on PVDF AND VDF/TrFE copolymer, respectively. Despite the responsivity of PVDF being slightly higher than that of the VDF/TrFE copolymer, the choice of which polymer to use will depend on the application and the preferred processing route.

A touchless passive infrared gesture sensor

A sensing device for a touchless, hand gesture, user interface based on an inexpensive passive infrared pyroelectric detector array is presented. The 2 x 2 element sensor responds to changing infrared radiation generated by hand movement over the array. The sensing range is from a few millimetres to tens of centimetres. The low power consumption (< 50 μW) enables the sensors use in mobile devices and in low energy applications. Detection rates of 77% have been demonstrated using a prototype system that differentiates the four main hand motion trajectories -- up, down, left and right. This device allows greater non-contact control capability without an increase in size, cost or power consumption over existing on/off devices.

A novel method for the performance control of a gas transmission compressor.

This paper presents the application of feed forward neural networks to the performance modeling of a gas transmission compressor. Results of compressor model testing suggest that compressor speed can be estimated to within ± 2.5 %. The neural network property of function approximation is used to predict compressor speed for given process constraints and instrument input sets. The effects of training set size, instrument noise, reduced input sets and extrapolation from the training domain, are quantified. Various neural network architectures and training schema were examined. The embedding of a neural network into an expert system is also discussed. A neural network can be retrained to reflect changing compressor characteristics. A global saving in compressor fuel gas of 1% could prevent the production of 6 million tonnes of CO2 per year, [1].

Teaching multitasking to electronics and computing students

Although widely used in the solution of engineering problems, multitasking is conceptually as complicated as parallel processing. A simple system that covers the key features of multitasking in a short time is presented. A description is given of its use with a final-year undergraduate class of electronics students who take computing options.

Comparison of human and machine recognition of everyday human actions

The research presented here makes a contribution to the understanding of the recognition of biological motion by comparing human recognition of a set of everyday gestures and motions with machine interpretation of the same dataset. Our reasoning is that analysis of any differences and/or correlations between the two could reveal insights into how humans themselves perceive motion and hint at the most important cues that artificial classifiers should be using to perform such a task. We captured biological motion data from human participants engaged in a number of everyday activities, such as walking, running and waving, and then built two artificial classifiers (a Finite State Machine and a multi-layer perceptron artificial neural network, ANN) which were capable of discriminating between activities. We then compared the accuracy of these classifiers with the abilities of a group of human observers to interpret the same activities when they were presented as moving light displays (MLDs). Our results suggest that machine recognition with ANNs is not only comparable to human levels of recognition but can exceed it in some instances.

A Comparison of Video and Infrared Based Tracking of Pedestrian Movements

We provide a comparison of the relative merits of video and infrared based methods for collecting pedestrian movements from the real world and also from experimental environments. We describe the underlying technological basis of both methods and the tools we have developed to help in collection and analysis of the data. The desire to collect such data is driven by the need of modellers and simulation packages to use base data that is founded in valid empirical evidence, rather than some form of inspired supposition, as is the case with many of the current systems. In addition to the collection of speeds we are also interested in understanding and quantifying the ranges of distances people deviate from a straight-on path when confronted by some obstruction in front of them.

A quantum car insurance paradox

I have just observed what might be a previously undocumented quantum mechanical phenomenon related to the EPR paradox so beautifully described by Cohn Jack (April p39). Last year I purchased a second-hand car, complete with a mechanical breakdown insurance policy. When I first read the policy document it seemed to cover most possible malfunctions, and I felt secure in the knowledge that I was unlikely to be saddled with any unexpected repair bills.