This paper and the following three papers were presented at the RIN97 Conference held in Oxford under the auspices of the Animal Navigation Special Interest Group, April 1997. The full proceedings, under the title Orientation and Navigation – Birds, Humans and Other Animals, can be obtained from the Director (£30 to Members, £50 to non-Members).

Studies of the compass mechanisms involved in the migratory orientation of birds have revealed a complex web of interactions, both during the development of orientation behaviour in young birds and in mature individuals exhibiting migratory activity. In young birds, the acquisition of compass orientation capabilities involves the interplay of apparently genetically programmed information with a suite of innate learning rules. The latter canalise the ways in which experience with relevant orientation information from the environment impinges on development. There are many general similarities with the development of singing behaviour in songbirds, although that system is more thoroughly understood, especially at the neuronal level.

Here we shall attempt to synthesise what is known about the development of compass mechanisms in a framework of innate information and learning rules. The way in which orientation behaviour develops leaves open the possibility for plasticity that enables birds to compensate for variability in the environmental cues that form the basis of their compasses. For at least some components of the system, behavioural plasticity remains into adulthood, allowing the bird on migration to respond in apparently adaptive ways to spatial and temporal variability in orientation information that it may encounter while enroute. We have studied these questions in the Savannah sparrow (Passerculus sandwichensis), a medium-distance North American emberizine nocturnal migrant. We will focus on that species, relating the results of our work to relevant studies on others.

The lives of sea turtles consist of a continuous series of migrations. As hatchlings, the turtles swim from their natal beaches into the open sea, often taking refuge in circular current systems (gyres) that serve as moving, open-ocean nursery grounds. The juveniles of many populations subsequently take up residence in coastal feeding areas that are located hundreds or thousands of kilometres from the beaches on which the turtles hatched; some juveniles also migrate between summer and winter habitats. As adults, turtles periodically leave their feeding grounds and migrate to breeding and nesting regions, after which many return to their own specific feeding sites. The itinerant lifestyle characteristic of most sea turtle species is thus inextricably linked to an ability to orient and navigate accurately across large expanses of seemingly featureless ocean.

In some sea turtle populations, migratory performance reaches extremes. The total distances certain green turtles (Chelonia mydas) and loggerheads (Caretta caretta) traverse over the span of their lifetimes exceed tens of thousands of kilometres, several times the diameter of the turtle's home ocean basin. Adult migrations between feeding and nesting habitats can require continuous swimming for periods of several weeks. In addition, the paths of migrating turtles often lead almost straight across the open ocean and directly to the destination, leaving little doubt that turtles can navigate to distant target sites with remarkable efficiency.

Dead reckoning (also called path integration) is the process by which a navigating organism derives its current position relative to an Earthbound reference point from its own locomotion. Dead reckoning requires the continuous estimation of changes in direction and location through self-generated signals and the computation of position on the basis of these signals.

(i) Hymenopterous insects measure rotations and translations mainly with the help of optical references such as the Sun and translational visual flow. By contrast, mammals are able to estimate their position on the basis of purely ‘internal’ information; that is, signals generated in the vestibular system by inertial forces, somatosensory feedback, and efference copies (copies of central commands that control the performance of rotations and translations). Obviously, the assessment of the angular and linear components of locomotion is much more precise if it is assisted by external references than if this is not the case.

(ii) Only man-made dead reckoning systems yield precise position information through the twofold integration over time of inertial signals deriving from angular and linear acceleration. On the biological level, all species tested so far seem to rely on a simplified form of path ‘integration’: in certain test situations, arthropods and mammals (including humans) commit similar systematic errors. This suggests that species from unrelated taxa update position according to a similar algorithm.

Although conditioning techniques are the most powerful way to study behavioural responses by animals to external stimuli, the magnetic sense has proved surprisingly resistant to conditioning approaches. This study demonstrated learned discrimination of magnetic field intensity stimuli by a new species, the rainbow trout (Oncorhynchus mykiss). In a unitary conditioned discrimination technique, four juvenile rainbow trout were trained to strike a target at the end of a response bar in anticipation of food. In successive experiments, the trout failed to discriminate the presence and absence of a vibration stimulus, but subsequently learned to discriminate the presence and absence of a magnetic field intensity anomaly (peak intensity of 75 μTesla). The authors conclude that the necessary conditions for training animals to magnetic intensity are the use of spatially distinctive stimuli and of a conditioned response that requires movement.

Of the great oceans of the world, the Atlantic, because of its violence, was the last to be mastered by man. The task in its entirety had to wait for the Portuguese sailors of the Renaissance. Isidore of Seville (c. 570–636), a Christian writer of the late Roman Empire, had written of the Atlantic that it was ‘incommensurable and uncrossable’. Although Pliny (a.d. 23–79) refers vaguely to the Canary Islands, all knowledge of them disappears in the Middle Ages until a Portuguese expedition under the command of the Italian Lanzarotto Malocello ‘re- discovered’ them in 1336. Italian charts of the XIVth century begin progressively to show the Canaries, Madeira, Porto Santo and the Azores, but all aligned along a N/S axis without any appreciation of the relative distances between them or how far they lay from the European shore. The first written evidence of the Portuguese ‘discovery’ of the islands of Madeira and Porto Santo appears in 1419–20 and of the Azores in 1427, about the same time as they began to be colonised under the aegis of Prince Henry of Portugal, called the ‘Navigator’. The difficulties of returning to them on regular voyages was to motivate the Portuguese to develop methods of measurement using the Pole Star as a navigational aid and this led, not only to a greater accuracy in placing the islands on the charts, but also to a greater precision in the charting of the west African coastline which they were progressively exploring during the second half of the XVth century.

Claims that Portuguese nautical astronomy originated in Aragon and was transmitted from there to Portugal or was introduced into Portugal from Germany by Regiomontanus and Martin Behaim have long ago been shown to be baseless.

The development of augmentation services, such as EGNOs and WAAS, requires careful modelling and simulation before any deployment of equipment takes place. The ideal location of ground stations, and the potential impact of temporary failures of ground equipment, must be assessed in detail, to ensure that proposed service levels of accuracy and integrity are maintained at all times. The correct analysis of these parameters can only be achieved if the various wide area processing models, and their respective error propagation characteristics, are suitably simulated. This paper describes the use of a sophisticated and versatile suite of GNSS analysis tools in recent studies. It describes the incorporation of realistic error budgets for WADGPS and LADGPS corrections, and the potential use of different combinations of GPS, GLONASS, geostationary and other satellites in both GNSS-1 and GNSS-2.

This paper is based on work done originally for the Chevron Shipping Company, whose agreement to its publication in its present form is gratefully acknowledged.

Ever since the International Regulations for Prevention of Collision at Sea (IRPCS) were introduced as a control system for collision avoidance, there have been disputes and arguments about exactly what they mean, and how they should have been applied in particular collision incidents. The principle of deciding by international agreement which ships should be given the duty of keeping clear of other ships generally works well but, as traffic volumes increase, so do situations in which the IRPCS alone do not provide a clear indication of what is expected of the mariner. Various solutions have been addressed to this problem, ranging from proposed radical revisions of the whole principle of collision avoidance, through periodic thoughtful adjustments to the details of the IRPCS, to informal additions and requirements imposed by individual shipowners on their fleets. The residual problem with any solution (except perhaps the first mentioned) is that there will always be exceptions where the rules are unclear or ambiguous. Adding adjustments may reduce these exceptions but, unless all uncertainties can be removed, the adjustments may create a new and even more complex set of exceptions, which are yet more difficult to interpret.

Accurate charts of tidal streams are needed in many fields of science and industry. The Proudman Oceanographic Laboratory's numerically modelled hydrodynamic data sets provide a suitable source for the production of such charts. Different methods of producing data in ‘tidal diamond’ format were investigated and the most suitable was selected for implementation over the UK continental shelf.

In recent years, both the offshore industry and the navies of the world have become increasingly interested in the potential operational usage of unmanned underwater vehicles. This paper provides a comprehensive review of a number of modern control approaches and artificial intelligence techniques which have been applied to the autopilot design problem for such craft.

This paper looks at the stimulus given by the practice and theory of navigation to certain problems in mathematical astronomy. The need for more accurate techniques of finding latitude and longitude, brought about largely by the great voyages of discovery and exploration as well as an increase in sea trade, gave rise to navigational instrument makers who produced devices of increasing accuracy. These craftsmen also made better measuring devices for the new observatories. Improvements in measurements led not only to new discoveries, but also made greater demands of theories underlying the practice of navigation. This gave impetus to the search for solutions to related problems in mathematical astronomy. Methods and special functions developed in this context were eventually to find application in a much wider range of problems in theoretical physics and engineering.

In this paper, two algorithms of Global Positioning System based attitude determination are proposed. The first algorithm extends the Kalman filter approach to determine the integer ambiguity and the orientation that is needed in a typical gps-based attitude determination problem. The second algorithm explores the mean field annealing neural network approach, which is a combination of the competitive Hopfield neural network and the stochastic simulated annealing technique, to resolve the optimal attitude problems. A test platform is set up for verifying these algorithms. The two algorithms are further compared in terms of computation speed and convergence rate.

I read Captain Gyldeń's account of failure of his GPS set (Vol. 50, 328, May 1997) with some interest. A few days earlier we had had a few thunderstorms while I had my personal hand-held GPS running, using its own built-in antenna and a 12 volt battery. It was lying on a bench inside a wooden hut where I keep some of my amateur radio equipment. After one particularly close flash and bang, which produced a one-inch spark from my transceiver antenna, the transceiver locked up and stopped responding to keyboard commands. Then I noticed the GPS set had also stopped working, showing only random symbols on its readout. I feared the worst but, after switching them both off, leaving them for a few minutes, and then back on again, they worked perfectly.

Microprocessors locking up in strong local electrostatic fields, perhaps? Maybe if I had simply left them alone they would have started working again after the charge had leaked away. Next time I nearly get hit by lightning I'll try it.

It was with the greatest interest that I read the two important papers by Dr N. Ward, published in the May edition of the Journal of Navigation.

The May 1997 issue of the Journal contains an essay by R. D. Pike on fast craft. I very much agree with his opinion on factual matters. However, I do not agree with the author's legal opinion on the existing COLREGS and I would like to suggest for consideration a much simpler solution in the future COLREGS.

In the May 1997 issue of the Journal, a paper on this subject was presented by Captain Turner on behalf of the General Lighthouse Authorities. I read the paper with interest, and in general concur with their assessment of the situation. However, I think that the situation regarding visual Aids to Navigation (AtoN) needs further consideration.

Many problems in navigation can be best viewed and solved as problems in analytical geometry. We only need to understand the geometry of two ‘navigable’ surfaces; the sphere and the ellipsoid of revolution. The ellipsoidal model is generated by revolving an ellipse about its minor axis and this model is used as a global model for the surface of the Earth. The eccentricity of the meridian ellipse is small (≈0·082) so we sometimes refer to this surface as a ‘spheroid’ since the surface is still ‘sphere-like’. The physical Earth is, in fact, referred to as a ‘geoid’ whose surface is that which approximates global mean sea level. The mathematical representation of the geoid is not trivial and the ellipsoid of revolution is an extremely good approximation to it.