GB1605386A - Improvements relating to vehicle guidance systems - Google Patents

Description

(54) IMPROVEMENTS RELATING TO VEHICLE GUIDANCESYSTEMS We EMI Limited, a British company of Blyth Road, Hayes, Middlesex do hereby declare the invention, for which we pray that a patent may be granted to us, and the method by which it is to be performed, to be particularly described in and by the following statement: This invention relates to vehicle guidance systems and has especial application to the automatic control of the flight path of aircraft.

Aircraft need to fly at low altitudes to remain invulnerable to enemy retaliation from the ground. In order to relieve the pilot of the strain of prolonged flight at low heights at high speed over undulating terrain, a range of radar sensors have been developed to provide this control automatically.

The simplest radars provide a "terrain following" facility in which the ground ahead is explored only in the vertical plane containing the flight path vector. More elaborate radars scan from side to side to provide "terrain avoidance" information so that the aircraft can steer in the horizontal plane in order to avoid going over exposed hills.

These radars are relatively bulky and expensive and have no other role to play when the aircraft is also equipped with forward looking infra-red sensory equipment hereinafter designated FLIR.

According to the present invention there is provided a guidance system for a moving vehicle including passive detecting means on said vehicle for successively receiving radiation incident thereon from areas of terrain relative to which the vehicle is moving, means for evaluating the distribution of said radiation over some at least of said areas, means for correlating the evaluated distributions relating to different ones of said areas to provide signals representative of relative displacement of said different areas and control means for utilising said signals to control the motion of said vehicle relative to said terrain.

The invention will be particularly described with reference to FLIR though a system using a low light television system (LLTV) could similarly be used.

Normally the aircraft will be flown only if the atmospheric conditions are clear enough for target detection by the FLIR and correspondingly the FLIR can be used for terrain following. The invention possesses the desirable feature that the terrain following equipment assumes a passive form rendering detection from the ground more difficult than with an active system such as radar.

There are known systems which use passive sensors for measuring the velocity-height ratio hereinafter referred to as v/h of an aircraft or missile. One arrangement is described and elaborated in U.S. Patents Nos. 2,878,711; 2,878,712; 2,878,713; 2,882,783 all to H.

Blackstone, 3,018,555 to F.G. Willey et al and also in an article "rR Velocity/Height Computer Studied" by B. Miller in Aviation Week, May 23, 1960. This uses two sensors with common optics arranged to receive signals from the ground vertically below the aircraft. The optics are arranged so that the two sensors "see" points on the ground which are displaced by a fixed angular distance in the along track direction. The output from one sensor is delayed until it corresponds to the output from the second sensor. Knowledge of the angular separation of the sensor images and the delay required for correlation between the sensor outputs enables v/h to be calculated. A different arrangement, described in U.S. Patent No. 3,076,095 to O.A. Becklund et al uses an IR tracker which scans the ground ahead until it locates a distinct IR source. This source is then tracked and its angular velocity at various angular positions is measured, thus giving a measure of v/h. The tracker uses a single detector with a rotating reticle to provide the tracking error signal which is used to steer the complete detector, reticle and optics assembly.

The present invention provides an arrangement which differs in concept from the known systems described in the immediately preceding paragraph in that it utilises an area correlation procedure to determine the angular rate of "streaming" of a patch of ground relative to an aircraft to enable the flight path to be controlled even in the absence of discrete targets.

Because area correlation, is used, the ground scanned is only required to provide some perceptible emission pattern within part of the area scanned for the autolock follow system to function. Thus, once it has been found in practice that such patterns usually exist, the invention enables known passive LLTV and infra-red scanning systems to be used for reliable automatic terrain following and avoidance without the need for an active scanning system, such as radar, which is sensitive to enemy countermeasures.

Aircraft in which the present invention may be employed are the interdictor travelling at 450 kts (750 ft/sec) at 200-400 ft and the helicopter moving at perhaps 90 kts (150 ft/sec) at a height of possibly 40 ft.

One embodiment of the invention uses the "elementary area" approach to the autolock follow technique as described in the specification of co-pending British Patent Application No. 30480/74 (Publication No.

1515295) to measure the angular rate of "streaming" at a specific angle of depression below the flight vector and to control the aircraft to keep this rate at a prescribed value.

In order that the invention may be clearly understood and readily carried into effect it will now be described, by way of example only, with reference to the accompanying drawings of which: Figure 1 is a diagram used to explain the geometric quantities used in this specification, Figure 2 is a vectorial representation of specific examples of certain quantities shown in Figure 1, Figure 3 is a block diagram of an example of the processing means according to an aspect of the invention, Figures 4 - 6 are diagrams used in the explanation of an aspect of the invention, and Figures 7 - 12 are diagrams used with reference to different examples of the invention.

Referring to Figure 1, and in terms of the parameters of this figure, assuming also that # < < 1

#o = v sin6ocose0=ve0 vûO R R whence R = v sin 0O cos 80 # v#o eo - -(3) and r = v sin 280 = v#o2 - (4) 0o e0 In the following only the approximate relations will be used for simplicity, but the exact expressions can be derived and instrumented without difficulty if desired.

The known "boresight" type of terrain following radar as employed for a 450 kt 750ft/sec flying at an altitude of 200 ft uses a value of the forward looking range R = 6000 ft.

Consequently, the depression angle required is 200 80=600o=33rad and the angular rate is 200 x 750 =4.2msad/sec (6000)2 Since the resolution of a typical FIlR is about 0.5 mrad. and as the time taken to estimate the angular velocity can be in the order of 1 sec and since the distance travelled in 1 sec is small compared with R, it follows that the instrumental accuracies in measuring both 8o and 0oare adequate.

In the case of the helicopter, the slower speed implies that for the same vertical acceleration the forward look range, R, can be reduced to about (6000/5X)ft =530 ft. Under these conditions 40 80=530=75m.rad and 40 x 150 #o = = 21 m.rad./sec (530)2 If in fact a time of 1 sec. is devoted to measuring velocity then in this time the distance travelled is 150 ft, a distance which is too large a fraction of R. But a measurement time of 05 sec. could prove acceptable, and this corresponds to detecting a change of 10.5 m.rads. These two situations are summarized in Figure 2 in which the flight vector FV is represented by and and tle magnitude of the angle equation is represented by + and the magnitude of the angle 80 is represented by the length of a dashed line.

Errors in measuring 6o and 80 lead to an error in r derived from eqn (4): 2 v #subo -v#o2 #r = #o ##o #o2 ##o 58o -r##o = 2 r #o #o or #r SO -##o -(5) r = 2 #o #o Consequently, the 0.5 m.rad. error in defining #o leads to negligible error in r. The error in 80 of about 0.5 m.rad/sec (in the more demanding case of the 450 kt aircraft) leads to proportional error of 0.5/4.2 = 0.12, corresponding to a height error of 24 ft in 200 ft. In the case of the helicopter, the height error is about 1.9 ft in 40 ft.

A particular example of a guidance system in accordance with the invention will now be described with reference to Figure 3. A sensor, 1, which in this particular example is a FLIR sensor senses the ground lying ahead of an aircraft on a viewing line extending towards the ground at an angle of80 with respect to the aircraft and scans the ground to build up an image of the ground as a matrix of points covering a field of view about the viewing line. The image of the matrix is displayed in a similar fashion to a television picture. Such techniques are well known and the same techniques used in this example could be applied equally to visual as well as IR images.

The requirement for the scanning is that a regularly updated image of the ground is available. A FIlR sensor for use in the invention employs a linear array of IR detector elements arranged so that the detectors provide images of adjacent patches on the ground angularity spaced in the along-track direction. Further optical techniques are used to scan the ground in the across-track direction, a suitable one being rotating prisms such as is described on Page 200 of Infra-Red System Engineering by R.D.

Hudson. In this example the system utilizes an optical arrangement to scan the ground in the track direction at the rate of 50 scans per second and 200 detectors to receive the scanned image and a final picture is produced by interlacing scans to give a frame rate of 25 frames second.

The line scans are sampled during the scan to give 512 samples per line so that the final picture consists of a matrix of 512 x 400 cells, produced at a rate of 25 frames/second. In cases where the resolution of the system in the along-track direction is limited by the detector spacing being too wide then the detectors can be displaced by half the spacing between adjacent detectors on alternate scans. Typically the system will have an angular resolution of 0.5 mrads. near the centre of the field of view with a total field of view of about 200 x 140.

It is important to know the orientation of the patch of ground being tracked relative to the aircraft and also relative to earlier views of the patch. Although it is assumed that normally the aircraft will be travelling in a straight line for at least the time over which the ground patch is being tracked, changes in drift angle and altitude have to be compensated for. These changes are monitored by the aircraft inertial navigation system and the compensation is performed as hereinafter described.

The signals from the sensor 1 are digitised in the analogue/,digital converter 2, a reference level for this converter being supplied by an AGC signal taken from a control computer 3. In this example data is available in serial form at a rate of 5.10 samples/sec. and the digitising is to eight bits. If necessary more bit digitization can be achieved at this rate and in certain cases one bit digitization may be sufficient.

An incoming picture store 4 is loaded with as much of the digital data as is likely to be used for terrain following, the loading being controlled by a control computer 3. Digital data representing target area or patch is selected when required and the digital values are placed in a target patch store 5. The suitability of a patch of ground as a target for tracking has to be tested before it is stored and tracking is initiated; this is because if the IR image shows little contrast or distinguishing pattern of different amplitude returns then it will be difficult to track. This testing is performed by a target patch selector unit 6 under the control of the control computer3 and the nature of the test can be varied according to the nature of the terrain. A simple test is to measure the mean amplitude of the digitized points and bearing in mind the limited dynamic range of the digitized signals if the mean amplitude value lies outside certain predetermined limits then the contrast is unlikely to be sufficient. If no suitable patches are available vertically below the flight vector, ground parallel to the flight direction but displaced to one side can be searched for suitable patches. When no patches at all are available for aprolonged period of time the aircraft is arranged to fly on its inertial navigation systems Once a target patch has been selected, data representing a further incoming picture is stored in the store 4. This further data from the store 4 is used to form patches to compare with the target patch in the target store 5. The choice of a patch is aided by assuming a value of e and estimating a new value of 0 for the target patch centre. The particular patch is chosen using these estimates by acomparison patch selection unit 7. The initial patch from the target store 5 and the new patches are compared in a correlator 8. This correlator is used to test for a similarity in pattern and amplitude between the two patches by comparing points in each pattern. Ambiguous readings are possible, especially for target patches with poor contrast or repeating IR patterns and a test of accuracy-of-fit such as is described in our co-pending patent application 30480/74 (Publication No. 1515295) is used. The results of the correlation are used to detennine the best estimates of 8 in unit 10 and average 60 in unit 11. Taking into account the aircraft motion a value of v/h can be computed in unit 12 and this can be used by an aircraft control unit 13 to maintain a constant value of v/h in a known fashion. Once the best estimate of the new patch position has been found a new picture can be taken from the sensor 1. The interval between picture samples will depend upon the time taken to complete the correlation process, although new pictures are available from a FLIR sensor at a rate of 25 per second.

The selection of suitable target patches and correlation patches will require data from the aircraft inertial navigation (I.N) system as shown in Figure 3. Data required includes the aircraft flight vector, angle rates and the altitude of the sensor 1. In certain circumstances it may be necessary to correct for the changing azimuthal displacement of target points with changing depression angle 8 as is more fully described below. The data values may require revising if the patch being tracked covers a wide azimuth angle; the use of interpolated values is one possible method. This correction is performed by a "Patch Geometry Adjustment" unit 9. The computer 3 maintains, as hereinbefore mentioned overall control of the system. It is used, for instance, to maintain the timing required to measure 8, organise the loading and reading of the data stores 4 and 5, to calculate the 'accuracy-of-fit' criteria, hereinbefore described, for the correlation process and target patch selection.

In a simple embodiment of the invention a patch to be tracked is shown in Figure 4, in which the flight vector FV is represented by a + and the magnitude of the angle 8o by the length of the dashed line. This patch comprises a square of 5 x 5 cells selected from the FLIR output centred on a viewing line extending at an angle 0O vertically below the flight vector.

A disadvantage of this system is that it covers only 25 cells (corresponding to a 2.5 mrad.

square), which is only just sufficient for reliable correction. An improved performance may be obtained by correlating over a larger solid angle such as shown in Figure 5, in which the flight vector and angle 8o are represented in the same way as in Figure 4. In this case, it may be necessary to allow for the change in geometry between successive samples of patches used to measure GO.

Thus consider an azimuthal slice only one cell wide in the elevation direction as shown in Figure 6 where it is presumed that the aircraft is flying at the prescribed height over a flat plane.

A cell at an azimuthal displacement of at the instant of the first sample will be at an azimuthal displacement of (1 + 0ou0o) at the second sample after time T. Consequently, for any postulated value of 8o against which a correlation is made, it is necessary to inject a small correction in order to compare corresponding cells despite the "streaming" effect. For large displacements from the vertical it is necessary to introduce more elaborate corrections.

By the foregoing means, it is possible to embrace a relatively wide strip of terrain. The limit is set by the extent to which the corresponding strip on the ground is horizontal in the direction normal to the line of sight. If a width of, say, 60 ft is taken for the interdictor case, the angular width is 60/6000, that is to say, 10 m.rad. or 20 cells. For the helicopter, a ground strip of about 25 ft may be assumed so that the angular width subtended is 25/500, namely 50 m.rad. or 100 cells.

The use of a single horizontal strip does little to increase the number of cells involved in the correlation. It is possible however to stack together a number of such strips; and to use their separate estimates of e to derive bO. But if the elevation coverage of the strips is relatively large then parabolic interpolation is necessary.

Alternatively, these "normalised rates" defined below may be averaged or interpolated. Since 8o is in the region of 33 or 75 m rad. it is practicable to cover an elevation span of about 10 or 20 m.rad. i.e. 20 or 40 cells respectively.

Consequently, the number of cells correlated in the case of the interdictor is 20 x 20 = 400, and in the case of the helicopter is 40 x 100 = 4000.

Bearing in mind that movement of any one cell occurs in only one prescribed direction, and that the maximum shift in the case of the helicopter is only 10 mead. or 20 cells between samples, it follows that the number of cell correlations required for each new measure of 80 is less than 20 x 4000, that is less than 80,000. Allowing 05 sec for each correlation and assuming that there are two sample sequences operating in "antiphase" it follows that the cell correlation rate required is 2 x 80,000/05 = 320/msec, which may readily be accomplished.

In another class of known "boresight" terrain following radar, a second probe is erected at a larger depression angle than the first and the outputs from the two probes are combined to provide an improved following characteristic over hilly terrain. The same facility can be achieved in accordance with the invention by correlating over a second patch as shown indicated in Figure 7. In a simple manner of combining the two outputs the aircraft is flown so as to keep the smaller of the height magnitudes rO and ri equal to the nominal clearance height.

Since v#o2 e0 = ro and 8i rl this procedure is equivalent to flying the aircraft so as to keep the larger of the two "normalised" angular rates o 2 and 82 equal to the required ratio v/r. There is no need for the two patches to be geometrically related exactly as indicated in Figure 7. They can be of different azimuthal and elevation widths if necessary.

The "twin boresight" system just referred to can be extended to any number of patches. In the limit, a relatively deep elevation patch may be provided, as shown in Figure 8 and the normalized angular rate of each elevation slice is then separately derived. In a simple embodiment, the largest of these values is used to control the aircraft. Such "individual" estimates are likely to be too noisy in practice however, and some local averaging or interpolation will be required before the normalised rates are compared.

Since the aircraft has time to manoeuvre to follow observed departres in r at long range, a better performance may be obtained by introducing the "ski-toe" characteristic used in known elevation - scanning terrain following radars. In this type of scheme, and as illustrated in Figure 9, both r and R need to be computed from the observed values of 80 at the corresponding values of 80 according to eqns (3) and (4). If the ski-toe law (derived from considerations of aircraft performance and the type of terrain to be encountered) defines, as a function of range, a clearance height S(R), then no terrain penetrates the ski-toe characteristic, if for every observed, and smoothed, rate Or at Or (v0,) Ve2 S # e, er The aircraft is then flown so that equality obtains between at least one pair of these sets of values at all times.

In all systems where more than one patch is tracked at one time alterations have to be made to the system as outlined with reference to Figure 3. These alterations entail a duplication of the target store 5, the correlator 8 and the comparison patch selection unit 7. The incoming picture store 4 provides the same basic data and the control computer 3 calculates the respective values of 8. Several patches can be tracked near the same value of8 and the values of vth obtained can be averaged. Also 8 can be measured by tracking patches over a wide range of 8. In the limit this can provide the data for a "ski-toe" terrain avoidance law as described above.

The interdictor aircraft requires high-quality navigation system for weapon delivery BO that correspondingly an accurate indication of the flight vector is available to the system of the invention as its reference position. In general a helicopter is not provided with such a facility.

In this case, it is possible to infer the flight vector position also from the FLIR output.

A simple arrangement to this end is shown in Figure 10. In this scheme, an auxiliary patch V is introduced "orthogonally" to that H used in the simple system of Figure 5. The patch H is used to derive 8o exactly as before except that, now, it cannot be assumed that there is zero lateral motion. Thus the system will correlate over a limited azimuthal range as well as over the elevation coverage required before. The elevation rate detected is used for terrain following as before while the azimuthal shift is an indication of the error in the azimuthal position of the assumed flight vector. This error is used to correct the flight vector position assumed in the next correlation cycle. In the case of the patch V, the observed elevation rate should be zero and any observed departure is used to correct the derived flight vector position in elevation.

An alternative arrangement is shown in Figure 11. In this case the auxiliary patch V is centred along the flight vector. Correlation of the patch with itself in both elevation and azimuth gives the errors in the assumed position directly.

Both of the above schemes rely upon correlation being performed on FLIR signals received from near the horizon, at least over flat ground, where the details often tend to be evanescent. Provided that the terrain is always approximately flat, this problem can be ameliorated by depressing the auxiliary patch V below the flight vector as shown in Figure 7 in which the + represents the flight vector and angles 8o, 81 are represented by dashed lines. In this arrangement, the lower patch is used for terrain following as before and also provides the observed error in the azimuthal positions of the inferred flight vector. The auxiliary patch V is also used to provide the observed elevation rate.

If there is an error o in the elevation position of the assumed flight vector and if the terrain is flat, then, from eqn (4), r (8o+ & ))2 (el +80)2 80 ego 80 whence 68=- 12 o 2(S8oOi - lbo) Bearing in mind that 8o and Ot are pre-fixed weights, and that 80 has to be driven to zero, a 8servo in the aircraft control unit 13 has simply to seek the situation in which bo b, 8o2 eft i.e., the situation in which the two normalised rates are equal.

In terrain avoidance the object is to enable the aircraft to go round hills rather than straight over them. It is necessary, therefore, to derive the altitude of the terrain over an angular sector ahead of the aircraft and centred on the flight vector.

For this purpose a relatively large number of patches such as that shown in Figure 12 are used.

Each patch provides the observed radial rate 8 from which the corresponding values of R and r can be derived. It follows, that the terrain in the patch is at a distance below the horizontal plane containing the flight vector of r sin 84 and displaced to the side of the flight vector by a distance r cos W. If the flight vector is pitched up by the angle a then the terrain in the patch is at an altitude below the aircraft of r sin W cos or - R sin a This value, together with the derived lateral displacement and R are all that is required to generate the conventional plane display showing as black the ground above a horizontal plane which is a preset distance below the aircraft and that below this plane as white.

In the vicinity of many of the cells being correlated there is likely to be few features for the system to "follow", either because the terrain is featureless in that locality or because the view is obliterated by local mist. In such cases it is important that spurious estimates of 6 are suppressed or at least deweighted.

A suitable weighting arrangement would be one of those described in the aforementioned specification.

In addition to forming the weighted sum of the elementary contributions to give er, the sum of the weights (i.e., the normalising factor) could be extracted as a measure of confidence. If this sum fell below a prescribed minimum, the aircraft would be automatically pulled-up away from the potential danger.

Another approach would be to use several adjacent bands similar to that in Figure 8 and arrange them to be contiguous. Normally, the strips other than the central one would be ignored but in the event that the terrain in that band was featureless the data derived from one or more pairs of the adjacent bands would be used instead, and providing that their integrity exceeded the threshold. The data from these outer strips would not be a-perfect measure of the clearance below the flight vector, and so under these conditions the nominal clearance height would be arranged to be greater than in the normal mode.

In regard to the interdictor, and the provision of the flight vector by the navigation system, the rms accuracy available is in the order of 0.3-5 m.rad. while that of the angle rate gyro is about 0.10/sec - 1.75 m.rad/sec.

The 5 m.rad. error in the flight vector position is acceptable for establishing the boresight, but in deriving the angular rate, as shown above the guidance system needs to operate to an accuracy in the order of 0.5 m.rad/sec. Consequently, it may be necessary to use a rather more accurate rate gyro than is commonly employed with radar terrain following systems.

For "single boresight" terrain following, it is proposed to measure oO at a prescribed depression angle ûO.

The alternative is to measure the value of 8o at which a prescribed value of 0o occurred. The aircraft would then be flown to drive the observed value of 80 to a prescribed value. This procedure requires the guidance system

Claims (9)

**WARNING** start of CLMS field may overlap end of DESC **. displaced to the side of the flight vector by a distance r cos W. If the flight vector is pitched up by the angle a then the terrain in the patch is at an altitude below the aircraft of r sin W cos or - R sin a This value, together with the derived lateral displacement and R are all that is required to generate the conventional plane display showing as black the ground above a horizontal plane which is a preset distance below the aircraft and that below this plane as white. In the vicinity of many of the cells being correlated there is likely to be few features for the system to "follow", either because the terrain is featureless in that locality or because the view is obliterated by local mist. In such cases it is important that spurious estimates of 6 are suppressed or at least deweighted. A suitable weighting arrangement would be one of those described in the aforementioned specification. In addition to forming the weighted sum of the elementary contributions to give er, the sum of the weights (i.e., the normalising factor) could be extracted as a measure of confidence. If this sum fell below a prescribed minimum, the aircraft would be automatically pulled-up away from the potential danger. Another approach would be to use several adjacent bands similar to that in Figure 8 and arrange them to be contiguous. Normally, the strips other than the central one would be ignored but in the event that the terrain in that band was featureless the data derived from one or more pairs of the adjacent bands would be used instead, and providing that their integrity exceeded the threshold. The data from these outer strips would not be a-perfect measure of the clearance below the flight vector, and so under these conditions the nominal clearance height would be arranged to be greater than in the normal mode. In regard to the interdictor, and the provision of the flight vector by the navigation system, the rms accuracy available is in the order of 0.3-5 m.rad. while that of the angle rate gyro is about 0.10/sec - 1.75 m.rad/sec. The 5 m.rad. error in the flight vector position is acceptable for establishing the boresight, but in deriving the angular rate, as shown above the guidance system needs to operate to an accuracy in the order of 0.5 m.rad/sec. Consequently, it may be necessary to use a rather more accurate rate gyro than is commonly employed with radar terrain following systems. For "single boresight" terrain following, it is proposed to measure oO at a prescribed depression angle ûO. The alternative is to measure the value of 8o at which a prescribed value of 0o occurred. The aircraft would then be flown to drive the observed value of 80 to a prescribed value. This procedure requires the guidance system to servo itself in elevation to locate and trackthe "velocity boresight". The procedure is applicable to the "twin boresight" solution also but, since a separate servo is required for each patch, it tends to become unwieldy for the more elaborate schemes. Because the patch can roam all over the field of view in elevation, it seems possible that the procedure provides a flying control that is less frequently interrupted due to the boresight pointing up into the sky when flying over hilly terrain. In the above description reference has been made solely to digital processing and storage. The invention is not, however, limited to such techniques and analogue processing systems comprising, for example, charge coupled devices can also be employed for data storage and signal processing in a manner which will be apparent to one skilled in the art. Additionally the above description refers to a FUR sensor but it is quite feasible to use visible radiation and a television camera instead. In conditions of poor lighting a low light level Television System (LLTV) can be used. In suitable conditions the use of colour picture provides an aid to tracking as well as better contrast for the autolock follow correlator. In general, however, IR systems are preferable in that they possess a better performance in fog and mist or hazy conditions and can be used in conditions when there is no visible radiation. Both IR and visible light techniques have the advantage over radar and similar active systems of using passive sensors and this renders them less vulnerable to detection and hence possible counter measures from an enemy. Referring back to the prior art mentioned hereinbefore, it will be seen that the present invention has certain advantages. Firstly more detectors are used and information from all the detectors and across the complete scan can be used if necessary. Tracking is undertaken on the basis of area correlation so that there is less likelihood of drop out due to indistinct signals. Discrete IR targets are not necessary and all that is required is a recognisable pattem of returns from the ground. This criterion can be satisfied in most environments even over apparently featureless terrain such as a desert which would not be amenable to optical autolock follow or discrete target tracking. Autolock follow tracking with FLIR can be extended to follow many targets or ground patches over a wide ground area and apart from terrain following it can, in principle, be used for terrain avoidance and flight vector derivation. WHAT WE CLAIM IS:

1. A guidance system for a moving vehicle including passive detecting means on said vehicle for successively receiving radiation incident thereon from areas of terrain relative to

which the vehicle is moving, means of evaluating the distribution of said radiation over some at least of said areas, means for correlating the evaluated distributions relating to different areas of said areas to provide signals representative of relative displacement of said different areas and control means for utilising said signals to control the motion of said vehicle relative to said terrain.

2. A system according to claim 1 wherein said detecting means comprises at least one forward-looking infra-red sensor arranged to detect infra-red radiation emitted from said areas of terrain.

3. A system according to, claim 1 comprising means arranged to examine said evaluated distributions relating to different ones of said areas so as to detennine the suitability of each of said areas for correlation.

4. A system according to claim 1 wherein said detecting means comprises a first detector arrangement to detect radiation from first and second areas consecutively and a second detector arrangement arranged to detect, synchronously with said first arrangement, radiation from third and fourth areas.

5. A system according to claim 4 wherein said first and second areas are orthogonal to said third and fourth areas.

6. A system according to any preceding claim wherein said vehicle is an aircraft.

7. A system according to claim 6 wherein the motion of the aircraft is controlled so that the height of said aircraft above said terrain is maintained substantially constant

8. A system according to claim 6 wherein the motion of the aircraft is controlled so that said aircraft maintains a substantially predetermined height above a reference level and wherein said control means is arranged to control said aircraft to perform substantially lateral manoeuvres so as to avoid obstructions detected by said detecting means.

9. A vehicle guidance system substantially as herein described with reference to the accompanying drawings.