When I presented my address to the Institute last year, I took as its theme ‘The Place of Navigation in Modern Air Transport’. Then I endeavoured to put before you some general thoughts and considerations, which I hoped might be stimulating to the further work of the Institute and its Members in the general field of air navigation.

The regulations governing visual avoidance of collision in the air and the carriage of aircraft lights were originally based on marine practice and extended to include the third dimension in aviation. These regulations were established first by the International Convention for Air Navigation and finally by the International Civil Aviation Organization.

There are three aspects of collision avoidance in the air which require mathematical investigation:

• (a) discovery and evaluation of the risk;

• (b) the remedial manœuvre;

• (c) the probabilities associated with (a) and (b).

One method of controlling aircraft is to assign each a given track, the distance between successive tracks being sufficiently great to ensure that the collision risk is very low. Most approaches to the problem of stipulating what this spacing should be, assume that the possible divergences of an aircraft about its preassigned path follow a Gaussian distribution with a standard deviation obtained by considering all the usual sources of navigational error (e.g. compass error and fixing error) together with meteorological errors due to inaccuracies in forecasting.

Apart from visual avoidance of other aircraft by pilots the only system of preventing air-to-air collisions at present is by means of air traffic control from the ground. With the increased speed of modern aircraft and increased traffic density on long-distance routes there is a tendency to demand further extension and development of air traffic control to solve the collision problem. The object of this paper is to examine the problems and limitations of air traffic control as a method of avoiding collisions between aircraft in cruising flight and how these are likely to be affected by future aircraft developments.

Vertical speeds of aircraft are always very slight compared with their horizontal speeds (except in the case of aerobatics) and it is first necessary to study the problem of air collisions in horizontal flight, a problem which, from the point of view of the geometry, is analogous to that of collisions at sea.

Statistics reveal that the vast majority of mid-air collisions in the past years have occurred during daytime under visual meteorological conditions in the vicinity of airports and at low altitudes. From the description of possible proximity warning or collision avoidance systems it follows that it is just this type of collision that causes technically the greatest difficulties for an operationally and economically feasible solution because of the multiple intruder problem, mutual interference, ground echoes, curved flight paths, and changes in aircraft attitude with resulting changes in aircraft speed. Under these conditions, the automatic pilot, the general introduction of which (for transport aircraft) has been recommended in the interest of collision avoidance, would be only of minor importance as it is common practice to switch off the automatic pilot whilst flying in the aerodrome traffic pattern.

The post-war development of air transport has temporarily out-paced the ability of the air traffic control service to provide for safe conduct which is always expeditious, and it has certainly out-moded the concept of visual avoidance. To alleviate the situation in the areas of high-density traffic in the United States, the Air Transport Association of America outlined a requirement for an airborne collision warning device.

Even in good weather, visual collision avoidance by pilots is unsatisfactory for reasons which include the following:

1. (i) When on a collision course the other aircraft has no relative motion across the windscreen and is indistinguishable from other marks on the windscreen until danger is very close.

2. (ii) There is a tendency to focus the eyes on nearby objects on the windscreen instead of at infinity, especially in the absence of conspicuous cloud and at high altitude. This can cause ‘empty field myopia’ whereby a normal pilot becomes nearsighted, his eyes focusing on a location less than six feet away.

3. (iii) In modern jet fighter aircraft the pilot must give much of his visual attention to instruments in the cockpit, such as the altimeter, owing to the difficulty of maintaining longitudinal stability, especially when heavily loaded. In clear air above 35,000 ft, it is estimated the pilot flies half contact and half on instruments.

For anti-collision purposes, the identification of an aircraft requires perception of its presence, detection of its track, and determination of its relative altitude. This evaluation will be recorded by the cerebral cortex and give rise to decisions regarding the action to be taken, which will vary from one individual to another. The avoidance manœuvre will be dictated by the situation and the rules of avoidance, and will involve a psychomotor coordination. Time is the essential factor in the chain of events, since the speed of the pilot's reactions will reduce the possibility of collision.

A thorough investigation is normally made of all the circumstances relating to an air accident. In-flight collisions can be of particular interest and should certainly be studied by doctors and psychologists. The human factor is said to play a part in 55 to 80 per cent of all accidents, and is basically psychological rather than physiological.

The usefulness of radar plotting, as at present practised, may be considerably enhanced by the application of a doppler radar technique which gives instantaneous measurement of relative velocity components and quickens the whole plotting process. The range and bearing of the target are given by an ordinary radar PPI, from which the rate of change of bearing may also be determined. The doppler radar gives an immediate measurement of the radial component of relative velocity and the derivative of radial velocity: from these data the course and speed of the target may readily be determined, using an automatic computer if so desired. Possible errors in deducing the true aspect and speed are discussed and the mode of operation, and principal design parameters of suitable doppler radar are given.

The conventional Decca Navigator system is of the type in which pairs of stations generate patterns of hyperbolic position-lines. One station, the master, is common to each pair so that a chain of three stations suffices to provide the two intersecting position-lines required for fixing. Among the advantages of this arrangement are the absence of any limit to the number of users of a single chain and the compactness of the position-fixing instrument (the receiver) which can readily be installed in the smallest craft. These two features contribute considerably to the success of Decca as an aid to surveying as well as to navigation.

When a resurvey of New Zealand coastal waters was started by the Royal New Zealand Navy, it was realized that many areas could not be covered accurately by visual fixing of landmarks and would require accurately laid beacons. It was also known that if accurate radar ranging could be provided, then a reduction could be made in the number of beacons to be laid and in the time required for the survey. As the standard Admiralty survey radar type 972M was not available as a complete equipment the Dominion Physical Laboratory were asked to develop a radar. This paper outlines some details of the radar and the operational results after two years installation on board H.M.N.Z.S. Lachlan during the survey of Hawke Bay.

The first amazement at hearing of the successful launching of the Russian Earthsatellite in October may have been followed in the reader's mind by some bewilderment at its rapid and seemingly erratic progress over all the world's capitals, large cities and outlying island groups. One supposed that a mathematical prediction would be possible but hesitated to attempt anything so complex. Yet, with some simplifying assumptions which must in any case be rather closely approximated by any near satellite that can continue to rotate in orbit, the problem is not difficult and lends itself to graphical treatment. The construction here proposed will give a good insight into the behaviour of such satellites and provide reasonably accurate predictions for a day or two ahead.