WO2001013138A1

WO2001013138A1 - Method and device at flying vehicle for detecting a collision risk

Info

Publication number: WO2001013138A1
Application number: PCT/SE2000/001566
Authority: WO
Inventors: Kristian Lundberg; Pontus BJÖRKLUNG; Niclas Appleby
Original assignee: Saab Transpondertech Ab
Priority date: 1999-08-12
Filing date: 2000-08-10
Publication date: 2001-02-22
Also published as: SE9902882D0; SE9902882L

Abstract

The invention presented here relates to a method employed by a flying vehicle (1) for detecting the risk of collision with at least one other flying vehicle (2). The method comprises collecting information on the position of at least one's own and the second flying vehicle in relation to the earth's surface, predicting, from the collected information, the subsequent course of one's own and the second flying vehicle for a predetermined prediction time, and deciding, from the predicted courses, if one's own flying vehicle (1) is at risk of colliding with the second flying vehicle (2), in which case a collision warning is issued and a manoeuvre for steering out of the collision course is indicated. If the proposed manoeuvre is not executed, the system perfoms the said manoeuvre. The invention relates also to a device in the flying vehicle designed to perform the above-mentioned procedure.

Description

Method and device at flying vehicle for detecting a collision risk .

TECHNICAL FIELD

The invention presented here relates to a method employed by a flying vehicle in the air for detecting the risk of collision with at least one other flying vehicle, as well as a device designed to detect the said collision risk.

STATE OF THE ART There are at present a number of position-finding systems used by aircraft, which, for example, employ radar and comprise a ground station.

US-A-5 872 526 relates to a position-finding system which does not require a ground station. Each aircraft is equipped with a GPS receiver. The aircraft transmits information on its own position, and receives and displays the corresponding position information from aircraft in the vicinity. As a result, it is possible for each aircraft pilot to manoeuvre their own aircraft so as to avoid collision in critical situations, at the same time as the pilots are communicating with each other to report their intentions as regards the course they are about to follow.

Systems for collision-risk detection, which attempt to prevent collisions-in the air, already exist. WO97/34276 describes a method for detecting collision risk in an aircraft. The method involves calculating the probability of one ' s own aircraft being present in predetermined sectors at a number of selected points in time/after which these probabilities for one ' s own aircraft and the probabilities for other objects are used to calculate the probability of one' s own aircraft, and at least one of the other objects, being present in any one of the sectors s i multaneously. DESCRIPTION OF THE INVENTION

One purpose of the present invention is to obtain, in relation to previously known techniques, an improved method for a flying vehicle to detect the risk of collision with other flying vehicles, and which works even for high-performance flying vehicles.

A further purpose of the present invention is to provide, in the event of collision risk, automatic manoeuvring without special communication between the flying vehicles.

This has been realised by a method of the type mentioned above, comprising the steps: a) collecting information on the current position of at least one' s own and the second flying vehicle in relation to the earth' s surface, b) predicting, from the collected information, the subsequent course of one' s own and the second flying vehicle for a predetermined prediction time and c) deciding, from the predicted courses, if one's own flying vehicle is at risk of colliding with the second flying vehicle. The method is characterised in that step a) involves recording the time for which the information was valid; step b) involves determining, for each flying vehicle, at least their current three-dimensional velocity, acceleration and acceleration derivative, either directly from collected information in a) or calculated from it, that mainly from the acceleration derivative, the variation in acceleration for the interval of prediction is calculated, that mainly from the current position, velocity and acceleration, as well as the calculated variation in the acceleration, each flying vehicle' s predicted acceleration, velocity and position is determined, for a number of prediction points in time within the prediction interval, so as to create a position vector, which gives the flying vehicle' s predicted course for the predicted time interval; and that step c) involves associating a course encompassing volume to each course and dividing the volume into volume elements where the flying vehicle is expected to be located at each point in the prediction time, whereby in the case of the volume elements from two or more flying vehicles at the same point in time being coincident or touching each other, the said collision risk is present.

The invention, in addition, comprises a device of the type mentioned in the introduction, which has the means for accomplishing the above mentioned features.

The invention-related method and device have several advantages over previously known techniques. A major advantage is that they are possible to use with high-performance flying vehicles, such as a fighter aircraft, since they use accurately time-stamped information updated at regular intervals. An additional advantage is that collision warnings are obtained according to two criteria, where the first gives a warning for increased preparedness if there is a risk of manoeuvring into a collision situation and the second criterium gives a collision warning where a direct collision risk is present and where it is necessary to take measures to manoeuvre out of the collision situation. A further advantage is that in the event of a collision warning, automatic evasive action is taken without special communication between the flying vehicles.

DESCRIPTION OF DRAWINGS Fig. 1 shows a collision risk situation involving two flying vehicles.

Fig. 2 shows a flow diagram of a first example of an inventive algorithm implemented in each aircraft.

Fig. 3 shows a flow diagram of a second example of an inventive algorithm implemented in each aircraft. Fig. 4 shows an aircraft and its predicted subsequent course and scope for manoeuvre.

Fig. 5 shows an example of a collision-detection device.

PREFERRED EMBODIMENTS

In Fig. 1 two airborne flying vehicles 1 ,2 are represented by, as an example, high-performance aircraft, which in general are military aircraft, which can perform advanced sorties with quick changes in direction.

In accordance with the first embodiment, shown in Fig. 2, each aircraft 1 ,2 gathers 3 information, accurately time stamped, on the position of its own and the second aircraft in relation to the earth' s surface. The position of one' s own aircraft is determined by, for example, using GPS, and the position of the nearby aircraft is determined via a computer link between one's own aircraft and this nearby aircraft, a so-called STDMA link for example. It is necessary for subsequent calculations that the position information for both aircraft is gathered at regular intervals, for example at a frequency of 10Hz, and that the information is accurately time- stamped. Based on the collected position information, the 3- dimensional velocities, accelerations, and acceleration derivatives of both aeroplanes can be estimated 4. The estimated velocities, accelerations, and acceleration derivatives, together with the positions comprise the input data for an algorithm for the calculation of a predicted course for each aeroplane for a predetermined prediction time, for which it is assumed that no new manoeuvres are executed during the prediction interval. A suitable prediction time for high-performance aircraft is, for example, 6-8 seconds.

In the algorithm, first the variation in acceleration during the prediction period is predicted 5, based on the calculated acceleration derivative, thereby creating a vector of acceleration from the estimated acceleration and its predicted variation. A vector of velocity is created 6 from the estimated velocity. Thereafter, the vector of acceleration is divided 7 into two components, one of which coincides with the calculated vector of velocity, the other being transverse to the vector of velocity in a plane, which is defined by the vector of acceleration and the vector of velocity; this division into components being made so that they can be later used in the detection of aeroplane manoeuvres, so called manoeuvre adaptation 8. The perpendicular component of acceleration is integrated 9 with respect to time in order to obtain a velocity surplus, which is added to the vector of velocity. In addition, the vector of velocity is compensated for sampling effects. Thereafter, the velocity surplus of the acceleration in a direction coinciding with the vector of velocity is summed with the compensated vector of velocity above. The vector of velocity is finally integrated 10 with respect to time in order to obtain a position in space. This method is repeated continuously for the entire prediction time for both aeroplanes. This results in a ground-related three- dimensional position for each point in time during the prediction interval, these positions together forming a position vector corresponding to the predicted trajectory of the aeroplane in question for the said prediction interval. To summarise, to achieve a good prediction of the position vector, an necessary condition is a high degree of accuracy of the collected information, which is accurately time-stamped and gathered at a relatively high frequency. Otherwise, there would be a high risk of large errors in the important acceleration derivative used in prediction of the aircraft' s subsequent course. An example of this is a position-finding system described in US-A-5 872 526, which yields sufficiently accurate position information.

In accordance with a second embodiment, shown in Fig. 3, one' s own aeroplane 1 has a navigation system 30, which supplies 1 1 information about the aircraft' s position relative to the surface of the earth, and its accelerations and rotations in a body fixed coordinate system. Furthermore, the second aeroplane 2 has a similar navigation system 30 and one's own aeroplane can access 1 1 corresponding information for the second aeroplane 2 via a data link. All information is accurately time-stamped. Based on the measured accelerations, the acceleration derivatives can be estimated 12 for both aircraft, so that together with the measured acceleration, they generate input data for an algorithm for the calculation of a predicted course for each aeroplane for a predetermined prediction time.

As in the previously described algorithm, the variation in acceleration during the prediction interval is first predicted 13 based on the calculated acceleration derivative. By using the knowledge of these aeroplanes, the courses of which are to be predicted, it is possible to enter boundaries in the algorithm (e.g. maximum values of acceleration in different directions) so that during the prediction interval, the physical limitations of the pilot or the aeroplane are never exceeded. In addition, a detection of aircraft manoeuv- res, a so-called manoeuvre operation, is performed 8. Thereafter, in each time-interval, a velocity vector is determined by integration 14 using the predicted acceleration in a body fixed coordinate system, together with the earth-related gravitation. First, however, the predicted acceleration has to be converted into a ground- related system of co-ordinates by predicting how the aircraft will be oriented at each time-interval during the prediction time. The orientation of the aircraft is predicted under the assumption that the calculated accelerations give rise exclusively to centripetal accelerations, and yields the curvature of the course in the hori- zontal and vertical plane; that is assuming that dynamic effects do not arise. Finally, the positions of the aircraft in a system of coordinates relative to the ground is calculated 15 for each time- interval during the prediction time by continually integrating the ground-related velocity of the aircraft with respect to time. Determined in this way, the position vector constitutes the coordinates of the aircraft' s predicted course in each time step. This procedure is repeated continuously for the entire prediction time for both aircraft.

The manoeuvre adaptation 8 method, mentioned previously, includes, apart from the method for limiting input parameters, such as velocity, acceleration, and acceleration derivatives, even methods for detecting aircraft manoeuvres. A manoeuvre is identified with the purpose of using it as a constraint in the prediction.

An example of manoeuvre adaptation is the case of different signs for acceleration and acceleration derivative at the moment of prediction. This can be interpreted as a manoeuvre (a change in velocity or a turn) that is about to be terminated. The acceleration is therefor set at zero if, during the prediction time, it changes sign. There is an exception to this rule, however: for acceleration changes in the direction transverse to the vector of velocity in a plane defined by the vector of acceleration and the vector of velocity, or normal to the aircraft' s horizontal plane, this relationship applies only for simultaneous change in the angle of roll. The acceleration is then permitted to change direction, which corresponds to the manoeuvre when the aircraft changes altitude and then levels off and remains at that altitude.

Another type of manoeuvre adaptation is achieved by checking if the aircraft' s angle of roll exceeds a predetermined value while its roll velocity operates with a sense of rotation which causes the angle of roll to increase, at the same time as, with an increasing acceleration in a direction transverse to the vector of velocity in a plane defined by the vector of acceleration and the vector of velocity, or normal to the aircraft' s horizontal plane, the prediction error can be reduced by making an intelligent guess that a turn of a certain load has been initiated. When the guessed load is reached by the aircraft, a new guess at another load is made, still on condition that both the angle of roll and the acceleration increase.

In this way, each turn performed by the aircraft is made discrete and a faster adaptation of the prediction is achieved.

Regardless of the method used to predict the course of the aircraft, transmitter error as well as uncertainties in the modelling are accounted for in the calculations, which means that for each aircraft a predicted most likely course 16, in addition to an uncertainty volume 17, preferably in the form of a cone, are obtained as illustrated in Fig. 4.

Predicted in this way, the courses with their uncertainty cones are used for collision detection, as will be described below.

In addition to the predicted course 16, a manoeuvre volume 18 is calculated for each aircraft, which preferably is mainly conical and which constitutes the volume that the aircraft can reach taking into account all possible manoeuvres during the current, previously mentioned, prediction time. In a refined version, knowledge of how the pilot normally handles the aircraft in different situations is used in the calculation. Furthermore, when calculating, the volume is given a statistical distribution, which gives the likelihood of the aircraft being present at different places in the manoeuvre volume. The perimeter of the manoeuvre volume symbolises the maximum physical performance of the aircraft and the pilot.

Using the manoeuvre volumes 18, it is decided if there is any risk that the aircraft, through active manoeuvring, will proceed into a collision situation, in which case a first warning call for increased preparedness is issued. This warning, it can be called a yellow warning, is indicated by means of illumination of, for example, a yellow lamp in the instrument panel.

If the estimated courses 16 of both the aircraft 1 and 2, together with the uncertainty volumes 17, also touch one another at one or more coincident points in time during the prediction interval, there is a risk of direct collision. In this case, a collision warning is issued. This warning, it can be called a red warning, is indicated by means of, for example, a red lamp connected to the instrument panel in the aircraft.

As is evident from Fig. 1 , in collision detection, each manoeuvre volume is divided into segments 19a, 19b, 19c, 20a, 20b, 20c, which represent each time-step in the prediction. The radial extension of these segments is determined by the maximum attained acceleration in the direction transverse to the vector of velocity in a plane defined by the vector of acceleration and the vector of velocity, or normal to the ground plane of the aircraft for each angle of roll, and its axial extension of maximum acceleration and retardation in the direction of the vector of velocity or the longitudinal direction of the aircraft. Accordingly, the possible positions of each aircraft are made known at each point in time. Each segment is supplied with a statistical distribution, e.g. according to a normal distribution curve, which indicates the probability of the aircraft being present in different areas along the length of the segment. If, at any point in time, both aircraft have segments that overlap each other, as, for example, the segments 19a and 20a in Fig. 1 overlap each other in the area 21 , there is at least a small risk that a collision could occur. If the area 21 lies in the predicted course of the aircraft, possibly with its uncertainty cone, a red warning is issued, i.e. a collision will most certainly occur if no active manoeuvre of the aircraft is performed. If, on the other hand, the area 21 lies somewhere else within each aircraft' s manoeuvre volume, in one embodiment a yellow warning is issued. In an alternative embodiment, the statistical distribution described above is employed, in which case the probability of each aircraft being present in area 21 is used to calculate a probability that both aircraft will be there. If this probability exceeds a selected pre-set first value, a yellow warning is issued, which means that if the current manoeuvre is continued, no accident will occur, but that there is a not insignificant risk of manoeuvring into a collision. A yellow warning is, therefore, a call for increased preparedness. If the probability exceeds a second, higher, selected pre-set value, a red warning is issued, as previously described.

The method described above for detecting collision risk naturally works equally well in situations where there are more than two aeroplanes.

All aeroplanes are equipped with systems designed to carry out the method described above. In one embodiment, the collision warning system is designed, in the event of a red warning, to give the pilot a specified pre-set time to manoeuvre out of the collision situation with the aid of a direction for evasive manoeuvre 28 recommended by the system. If the pilot fails to take evasive action, the system performs this instead. An important feature of each aircraft system is that it is designed to take evasive action without any further communication between the aircraft, and nevertheless ensure that the aircraft pass each other at a safe distance. The system-generated evasive manoeuvre is based on all aircraft having the same input data and using the same algorithm for calculating the evasive manoeuvre. Each aircraft is thus supplied with a unique direction so as to avoid collision. In an example with two aircraft involved, the cross product between a combination of each aircraft' s normal vector and vector of velocity is used in order to generate an evasive manoeuvre. For a situation involving two or more aircraft a simulation is made, in a calculating unit for example, of a manoeuvre combination where each aircraft involved performs a defined manoeuvre. The manoeuvre consists of an initial rolling motion to a predetermined roll angle, as well as a defined phase of acceleration. The manoeuvre combination is changed by increasing the pre-set roll angle of one of the aircraft in the initial rolling motion. When all combinations of initial roll angle have been simulated, one chooses the combination that provides the greatest minimum distance between the different aircraft.

The direction of flight that each aircraft is recommended to perform according to one of the above methods is displayed for the respective pilots on the cockpit display device 22 in the aircraft.

The collision warning system described herein can naturally be combined with a ground collision warning system, which warns the pilot if he/she is approaching ground at a too high speed.

As is evident from Fig. 5 , the collision warning system 5 comprises means for gathering information on at least one' s own position, for example in the form of a GNSS receiver 25 or another navigation system 30, and means 23 for gathering corresponding information from nearby aircraft, for example in the form of a

STDMA link. The gathered information is treated by a calculating device 24, for example incorporated in software, which controls a display device 22 with a red indicator 26, a yellow indicator 27, in addition to a direction indicator 28, and a manoeuvring device 29 designed to steer the aircraft out of a collision situation.

Claims

1 A method employed by a flying vehicle ( 1 ) for detecting the risk of collision with at least one other flying vehicle (2), the method comprising the steps:

a) collecting information (3 ; 1 1 ) on at least the current position of one' s own and the second flying vehicle in relation to the earth' s surface, b) predicting, from the collected information, the subsequent course ( 16) (4-9; 12- 15) of one' s own and the second flying vehicle for a predetermined prediction interval and c) deciding, from the predicted courses ( 16), if one' s own flying vehicle ( 1 ) is at risk of colliding with the second flying vehicle (2), characterised in that a) involves recording the time for which the information was valid, b) involves determining, for each flying vehicle, at least their current three-dimensional velocity, acceleration, acceleration derivative, either directly from collected information in a) or calculated from it, that mainly from the acceleration derivative, the variation in acceleration for the interval of prediction is calculated, that from the current position, velocity and acceleration, as well as the calculated variation in the acceleration, each flying vehicle' s predicted acceleration, velocity and position is determined, for a number of prediction points in time within the prediction interval, so as to create a position vector, which gives the flying vehicle' s predicted course for the predicted time interval and that c) involves associating a course-encompassing volume ( 17, 18) to each course and dividing the volume into volume elements where the flying vehicle is expected to be located at each point of time in the prediction interval, whereby in the case of the volume elements from two or more flying vehicles at the same point in time being coincident or touching each other, the said collision risk is present.

Method according to claim 1 , characterised in that c) involves dividing each course-encompassing volume into a uncertainty cone ( 17) representing uncertainties resulting from, for example, transmitter errors, and a manoeuvre volume ( 18) encompassing the uncertainty cone ( 17), which manoeuvre volume represents the volume that is possible to reach in the prediction interval with predetermined manoeuvrability for each flying vehicle, that a collision warning (26) is issued in the case of the volume elements from the said uncertainty cones ( 17) coinciding or touching each other at the same points in time, and that a warning call for increased preparedness (27) is issued if the manoeuvre volumes ( 18) coincide or touch each other at the same points in time .

Method according to claim 1 , characterised in that after the issue of the collision warning (26) an evasive manoeuvre (29) is performed to steer out of the collision course.

A device in an flying vehicle ( 1 ) arranged for detecting, in the air, a collision risk with at least one other flying vehicle (2), said device having means ( 19, 20) arranged to collect information on at least the position of one' s own ( 1 ) and the second (2) flying vehicle in relation to the earth' s surface; calculating means (21 ), communicating with the information- collecting means and arranged to predict, from the collected information, the subsequent course ( 16) of one' s own and the second flying vehicle for a predetermined prediction interval, and to decide, from the predicted courses ( 16), if one' s own flying vehicle ( 1 ) is at risk of colliding with the second flying vehicle (2); as well as a display device (22), communicating with said calculating means and arranged to indicate that a risk of collision has arisen, characterised in

that said information-collecting means ( 19, 20) are further arranged to couple, to the collected information, a time for which the information was valid, and

that said calculating means (21 ) are arranged to determine, for the each flying vehicle either directly from the collected information or calculated from the same, at least the flying vehicle' s current three-dimensional velocity, acceleration, and acceleration derivative; to calculate, mainly from the current acceleration derivative, the variation in acceleration for the interval of prediction; to calculate, mainly from the current position, velocity and acceleration, as well as from the calculated variation in the acceleration for a number of points in time within the prediction interval, the flying vehicle' s predicted acceleration, velocity and position in order to create a position vector, which gives the flying vehicle' s predicted course (16); to associate a course-encompassing volume ( 17, 18) to each course and divide the course into volume elements where the flying vehicle is expected to be located at each point of time in the prediction interval, whereby in the case of the volume elements from two or more flying vehicles at the same point in time being coincident or touching each other, the said collision risk is present.

A device according to claim 4, characterised in that said calculating means (21 ) are arranged to divide the course encompassing volume ( 17, 18) into an uncertainty cone ( 17) representing uncertainties resulting from, for example, transmitter errors, and a manoeuvre volume ( 18) encompassing the uncertainty cone ( 17) and representing the volume that is possible to reach in the prediction interval with predetermined manoeuvrability for each flying vehicle, and that the display device (22) is arranged to issue a collision warning in the case where the volume elements from the said uncertainty cones coincide or touch each other at the same points in time, and to issue a warning call for increased preparedness (27) if the manoeuvre volumes ( 18) coincide or touch each other at the same points in time.

A device according to claim 5, characterised in that it has manoeuvring means (23) arranged to perform, on issue of the collision warning, an evasive manoeuvre to steer out of the collision course.