Patent Application
20140320640
The field of the invention is that of optical devices making it possible to measure the orientation of an object in space without contact. There are various possible fields of application, but the main application is that of detecting the posture of the helmet of an aircraft pilot, thus making it possible to project an image into his visor in exact superposition on the exterior landscape or to slave various systems of the aircraft under his view. The precision sought in such systems is of the order of one milliradian.
There are various optical techniques for measuring the orientation of a helmet. Generally, conspicuous elements are installed on the helmet, and these are located by a system of cameras. The positions of the images of these conspicuous elements make it possible to determine the orientation of the helmet by calculation.
These elements may be passive or active. Passive elements are illuminated by an external source. To this end, retroreflective corner cubes or retroreflectors may be used. It is sufficient to arrange the optical emission and reception components on the same axis.
These systems with retroreflectors have low sensitivity to sunlight. They are combined with one of the following types of fixed devices:
In the latter two arrangements, the reflector is equipped with a mask which is transmissive in the central part and opaque at the periphery, and which is applied onto its entry face. The contour of the mask is in the shape of a parallelogram, thus embodying the orientation of two fixed directions of the helmet. The orientation of the helmet is calculated by analysing the shape of the contour projected onto the sensor. The analysis relates to the transitions between the light and dark regions of the reflection received by the sensor.
The latter arrangement leads to an optical device which is simple and has a long depth of field. For an elementary corner cube, however, the angular detection field remains limited for the following reasons:
By way of example, a corner cube reflector CC is represented in
The angular field of this reflector is limited by the shape of the reflector and by that of the mask, as can be seen in
The maximum angular field of the reflector in incidence or in reflection is therefore given by the three reflective planes POQ, QOR and POR. The various values of this angular field can be calculated as a function of the impact point of the central ray passing through the vertex O on the mask. As a first example, for a mask corresponding to the parallelogram ABCP of
These field limits are significantly far from those corresponding to the angular extent desired around the diagonal OIJ in the half-space bounded by the plane PQR.
The system of the invention overcomes this deficiency. The solid angle obtained is that defined by the vertex O and the entire triangle PQR, that is to say by the three planes of the trirectangular trihedron OXYZ, its value being π/2 steradians.
In order to obtain a large angular field, the system according to the invention comprises corner cubes having the following two original characteristics:
Compact combinations of reflectors according to the invention make it possible to ensure measurement of the position detection in a large angular extent.
Furthermore, the analysis method is sensitive neither to luminous power variation of the source nor to variation in its colorimetry, nor source/reflector distance variation due to the position of the helmet.
More precisely, the invention relates to a system for detecting the posture of an object which is mobile in space, comprising a fixed electro-optical device of known orientation comprising at least a first emission source, image analysis means and an optical assembly comprising at least one optical corner cube arranged on the mobile object, characterized in that:
Advantageously, the entry face of the corner cube being triangular, it is divided into three identical coloured triangular regions.
Advantageously, the entry face of the corner cube been triangular, it has three identical coloured triangular regions enclosing a transparent triangular central region without a spectral transmission filter.
Advantageously, the specific markings are geometrical and/or photometric markings.
Advantageously, in a first embodiment, the optical assembly has four adjacent tetrahedral corner cubes of identical shapes, each corner cube having one entry face, two reflective lateral faces common to two other corner cubes, and a reflective third lateral face located in a plane common to the other third lateral faces of the three other corner cubes, the said third lateral faces thus forming a square, the four entry faces and the said square forming a pentahedron in the form of a square-based pyramid. In an alternative arrangement, the optical assembly has four tetrahedral corner cubes of identical shapes arranged symmetrically around a single vertex common to the four corner cubes, each corner cube having one entry face. In a first variant, the entry faces of the first and second corner cubes comprise the same first triplet of coloured regions, and the entry faces of the third and fourth corner cubes comprise the same second triplet of coloured regions, which is different from the first triplet of coloured regions. The specific marking of the sides of the coloured regions of each entry face is arranged in such a way that this face can be distinguished from the three others. In a second variant, the entry faces of the four corner cubes comprise the same first triplet of coloured regions and the entry faces of the first and second corner cubes have a neutral filter of predetermined transmission.
Advantageously, in a second embodiment, the optical assembly has eight adjacent tetrahedral corner cubes of identical shapes, each corner cube having one entry face, and three reflective lateral faces common to three other corner cubes, the eight entry faces forming a regular octahedron. In a first variant, the entry faces of the eight corner cubes comprise the same first triplet of coloured regions and the entry faces of the four corner cubes have a neutral filter of predetermined transmission. In a second variant, the entry faces of the first, second, third and fourth corner cubes comprise the same first triplet of coloured regions, and the entry faces of the fifth, sixth, seventh and eighth corner cubes comprise the same second triplet of coloured regions, which is different from the first triplet of coloured regions. The specific marking of the sides of the coloured regions of each entry face is arranged so that this face can be distinguished from the seven others.
Advantageously, in a third embodiment the optical assembly has a plurality of corner cubes, the entry faces of which are oriented and positioned along the pentagonal faces of a regular dodecahedron or of a part of a dodecahedron, each entry face being oriented in each pentagonal face so that the optical assembly produces a reflection on the detector in a range of determined orientation, the specific marking of the sides of the coloured regions of each entry face being arranged so that this entry face can be distinguished from all the others.
Advantageously, the fixed electro-optical device has at least one point emission source and only optical components having a zero or quasi-zero optical power, that is to say plane mirrors or semi-reflective plane plates, the separation between the point emission source and the light reflected by the corner cube or cubes been produced by means of a semi-reflective plane plate.
Advantageously, the analysis means comprise electronic preprocessing arranged at the output of the matrix sensors and making it possible to assign a three-component code to each pixel, each component being representative of a predetermined spectral band, each component being encoded over a limited number of levels representing the absence or presence of light received in the said spectral band.
The invention also relates to a flight helmet comprising at least one optical corner cube, the entry face of which is divided into at least three separate coloured regions, each coloured region having a spectral transmission filter, each filter transmitting only a predetermined spectral band different from those of the other spectral transmission filters, each side of the said coloured regions having a specific marking making it possible to identify this said side, the said corner cube being intended to operate in a system for detecting the posture of a mobile object as defined above.
Advantageously, in a first variant, the helmet comprises at least one regular dodecahedron having a plurality of identical pentagonal faces, each having a corner cube arranged so that the entry face of the said corner cube is located on one of the said pentagonal faces and each entry face is oriented in each pentagonal face so that the optical assembly produces a reflection on the detector in a range of determined orientation, the specific marking of the sides of the coloured regions of each entry face being arranged so that this entry face can be distinguished from all the others.
Advantageously, in a second variant, the helmet comprises at least two identical regular semi-dodecahedra, each having six identical pentagonal faces, each having a corner cube arranged so that the entry face of the said corner cube is located on one of the said pentagonal faces and each entry face is oriented in each pentagonal face so that the optical assembly produces a reflection on the detector in a range of determined orientation, the specific marking of the sides of the coloured regions of each entry face being arranged so that this entry face can be distinguished from all the others.
The invention will be understood more clearly, and other advantages will become apparent, on reading the following description which is given without implying any limitation, and by virtue of the appended figures, in which:
In what follows, a first part of an optical detection system according to the invention comprising a single reflective corner cube will be dealt with. A second part deals with a detection system comprising an optical assembly comprising a plurality of associated corner cubes.
Part One: Detection System Unique with a Single Corner Cube
The detection system for optical detection of the posture of an object which is mobile in space according to the invention comprises a fixed electro-optical device of known orientation, image analysis means, and an optical assembly comprising at least one polychromatic optical corner cube arranged on the mobile object. The core of the device is the polychromatic corner cube. This corner cube can operate with various electro-optical devices. Notably, mention may be made of devices comprising white light emission sources and colour reception cameras. However, it is particularly suitable for an electro-optical device with a point white light source.
An example of this type of device is represented in
In the rest of the description, the component CCC will be referred to arbitrarily as a corner cube or reflector. The corner cube is secured to the mobile object. In aeronautical applications, the mobile object is a helmet. The corner cube may be a solid reflector, in which case the reflection on the internal faces of the corner cube takes place by total internal reflection. It may also be formed by assembling three plane mirrors arranged orthogonally to one another. In the rest of the description, unless otherwise specified, the corner cubes may equally well be solid or not.
The device according to the invention can also operate with electro-optical devices having only a single matrix sensor.
The entry face of the corner cube CCC is divided into at least three separate coloured regions each coloured region having a spectral transmission filter. By way of example, an entry face according to the invention is represented in
In what follows, the following colorimetry conventions have been adopted. The visible spectrum is divided into three large spectral bands referred to as “red”, “green” and “blue”, from the longest wavelengths to the shortest wavelengths. A red-coloured filter transmits only the red spectral band and filters out the green and blue spectral bands. The complementary colours, referred to as “magenta”, “cyan” and “yellow” have as respective spectral bands:
The spectral positioning and the width of the bands are, of course, adapted according to the spectral sensitivity of the matrix sensors. It is possible to adapt the detection system for operation in other spectral distributions, such as the near infrared. It is sufficient to keep the three different spectral bands and to adapt the filters and the sensitivity of the sensor accordingly.
Analysis of the coloured images coming from the matrix sensors, and in particular their vanishing lines, makes it possible to find the position and the orientation of the corner cube CCC.
As mentioned, the particular feature of the corner cube according to the invention is that it has a polychromatic front face. Various arrangements of the coloured regions of this front face exist. In a first embodiment, illustrated in
With this arrangement of the entry face, the following effect is obtained:
In the image projected onto each sensor, only red, green, blue or white surfaces of homogeneous colour are therefore found, these surfaces not being mixed and having no composite colour obtained by superposition. These surfaces are adjacent and have no intermediate black regions. In the rest of the description, a “colour dominant” surface denotes the surface resulting from the juxtaposition of a surface of pure colour, either red, green or blue, and a white surface. “Red dominant”, “green dominant” or “blue dominant” surfaces are therefore obtained. At the pixels of the sensors, these “colour dominant” surfaces are characterized as follows:
The contour of each of these three colour dominant surfaces is a parallelogram if the source is at infinity, or a quadrilateral in the general case of a source at a finite distance. Each surface is produced by:
More precisely,
The shape of the contour and the colour of surfaces obtained on the sensors depend on the incidence of the radiation produced by the source S on the reflector, and more precisely on the position of the point T of intersection on the plane PQR of the straight line OS joining the source S and the vertex O of the reflector. For example, the image on the detector contains red if and only if the point T lies inside the rhombus ABCP.
By way of examples,
In
By carrying out projection into the two planes of the sensors CM1 and CM2 of the images reflected by the corner cubes, these planes being different and of known positions and orientations, two different quadrilaterals are obtained. From knowledge of the position on each sensor of the two vanishing points of the same projected rhombus, the position of the vertex O of the reflector and the orientation of two of the adjacent sides of the corresponding initial rhombus are deduced. The orientation of the position of the mobile object is thus obtained.
So long as the orientation of the reflector is such that the intersection T between the straight line SO joining the source S to the vertex O of the reflector and the plane PQR remains inside the triangle PQR, there is at least one quadrilateral imaged onto the detectors. This quadrilateral is:
This arrangement allows a significant increase in the angular detection field in so far as all of the entry surface of the corner cube, and no longer a partial mask, is now taken into account in the detection. For a corner cube according to the invention, the total angular field obtained is π/2 steradians.
If the point source is not at infinity, the parallelograms of the dominant coloured surfaces are deformed into quadrilaterals. The point T′ remains the circumcentre common to the segments P′0P′, A′0A′, B′0B′, C′0C′, but without being at the centre.
The same filtering method is applicable, by means of simple adjustments, with the same mosaic image sensors by using a reflector provided with less spectrally selective filters. For example, yellow, cyan and magenta dominant filters may be used. The projected quadrilaterals are also yellow, cyan and magenta dominant.
If simple coloured triangles are used, it is difficult to determine the sides of the triangle involved in the measurement. In order to distinguish the sides, a specific marking is added to them in the proximity of each vertex of the triangles PAC, RAB and QBC. On the projected image contour of a given dominant colour, the original vertex and one of the two sides converging on this vertex are thus identified. Various types of geometrical, photometric or colorimetric marking may be used. For example, small local maskings of different shape may be produced.
The same marking as that of the triangle APC is carried out for the two other triangles RAB and BCQ, respectively green and blue.
For the projected images which do not have marking, such as the luminous dodecagon of
In a second embodiment, illustrated in
The luminous surface obtained on the detector, for a source S at infinity, varies according to the position of the point T defined as above and positioned, for example, inside the magenta triangle PIQ. It is represented in
For the first position of T close to the point Q, the surface represented in
For the second position of T close to the point I, the surface represented in
Part Two: Detection System Comprising a Plurality of Corner Cubes
Even though an optical detection system having a single coloured corner cube according to the invention has an angular field greater than that of a system having a corner cube with a parallelogram contour mask, it may be insufficient in a certain number of applications which require a large angular range. This is the case, notably, with helmet posture detection systems. The simplest procedure is to arrange an optical assembly, having a plurality of corner cubes arranged so as to cover a wide angular sector, on the mobile object. This arrangement also has the advantage that, under certain conditions, the system can function with a single emission and reception device.
The combination of a plurality of corner cubes is known. For instance, U.S. Pat. No. 6,123,427 entitled “Arrangement for retroreflection of a ray using triple prisms” describes a plurality of optical combinations having from six to ten reflectors. U.S. Pat. No. 3,039,093 entitled “Reflective radar target” describes an arrangement having twenty reflectors. Lastly, patent FR 78 24013 entitled “Dispositif optoléctronique de detection et de localisation d′objet et système de repérage spatial de direction comportant un tel dispositif” [Optoelectronic device for object detection and localization and spatial direction identification system comprising such a device] describes, for applications in the same technical field as the invention, a combination of four corner cubes which are adjacent along two of their reflective faces so as to widen the angular field.
These various combinations, however have several significant drawbacks for our application, which are detailed below:
The corner cube combinations described below all have corner cubes whose triangular entry face has three identical coloured triangular regions surrounding a transparent central triangular region. An example of this type of entry face is represented in
A first advantageous corner cube combination consists in combining four adjacent corner cubes denoted Re1, Re2, Re3 and Re4, such as those described in
It is necessary to distinguish the reflectors from one another. One possible way of distinguishing them is obtained by combining corner cubes with different polychromatic filters. For example, the reflectors Re1 and Re3 are equipped with red, green and blue triangular filters and a neutral central triangle. The reflectors Re2 and Re4 are equipped with cyan, magenta and yellow triangular filters and a neutral central triangle.
The reflections generated by the corner cubes Re1 and Re3 are distinguished from those of the corner cubes Re2 and Re4 by the absence of the three colours cyan, magenta and yellow in the projected image. The reflections generated by Re2 and Re4 are distinguished from those of Re1 and Re3 by the presence of at least one of the three colours cyan, magenta and yellow in the projected image.
In order to identify the source reflector from the isolated reflection of a given colour, that is to say in order to choose between the two corner cubes Re1 and Re3 or between the corner cubes Re2 and Re4, additional differentiation is added on the markings of the faces of the corner cubes. As examples, this additional differentiation is obtained:
For the projected images which do not contain marking, the differentiation is carried out by the sense of the sequencing of the colours surrounding the white transparent surface. Thus, on the image provided by Re1, the sequence red, green, blue surrounding the transparent surface follows the anticlockwise sense, whereas it follows the clockwise sense for Re3. On the image provided by Re2, the sequence magenta-yellow-cyan surrounding the white surface follows the anticlockwise sense, whereas it follows the clockwise sense for Re4.
For a given reflector the existence of a reflection is ensured so long as there is a vertex/source ray SO inside the trihedron formed by its three reflective faces. In order to ensure on the one hand the existence and on the other hand a minimum dimension of the reflection when the source is in the vicinity of one of the two planes XOY and XOZ, the configuration of the four corner cubes is modified. Relative to the arrangement of
In this case, the source S must be positioned at a minimum distance xmin from the reflectors in order to ensure continuity of angular coverage from one reflector to another. It can be shown that:
xmin=e/[2tg(α/2)], e being the maximum distance separating two vertices of two adjacent corner cubes and a being the angle of inclination existing between the two adjacent faces of these two corner cubes.
When the source S is in the vicinity of one of the two planes XOZ and XOY, and only in this case, the two adjacent reflectors each generate an image on the detector, these two images therefore being partially superposed. Where there is superposition, the light powers projected onto the sensors are added together. Consequently, and in this particular case, the processing of the images coming from the sensors must be capable not only of recognizing the colours of the coloured areas contained in the captured images, but also their amplitude level. The simplest procedure is to arrange a neutral filter of known attenuation on some corner cubes. For example, if the choice is made to attenuate the light levels reflected by the corner cubes Re2 and Re4 by a factor of two relative to those of the corner cubes Re1 and Re3, a global filter of transmission (0.5)0.5 is added on the entry faces of the corner cubes Re2 and Re4.
The choice and the arrangement of the colours on all the four reflectors make it possible to recover the two original images on the basis of a combined image.
In this configuration, it is not necessary to determine the luminance levels in the captured images precisely, but instead the relative amplitude levels of the different coloured areas with respect to one another. Thus, each coloured pixel may be assigned a simple three-digit code which depends on predetermined thresholds.
For example, the first digit represents the colour red, the second digit the colour green and the third the colour blue.
Each digit has at most five values. For example, the value 0 indicates that the colour is entirely absent from the pixel. The value 1 indicates that the colour comes from one and only one attenuated coloured area. The value 2 indicates that the colour comes from one and only one unattenuated coloured area. The value 3 indicates that the colour comes from two areas, one attenuated and the other not attenuated, and so on.
Thus, a pixel belonging to the sum of the images of the three reflectors or of the four reflectors has the following code:
The letters r, j, v and c denote the colours red, yellow, green and cyan of the images coming separately from the reflectors.
With knowledge of the codes of the surfaces of the total image by measurement, the surfaces of each of the two to four constituent images can therefore be reconstructed using their code, i.e.:
The contour of the surfaces of uncombined colours produced separately by the reflectors, and therefore the orientation of at least one of these reflectors, is deduced therefrom.
This method has two significant advantages. On the one hand, it does not employ an absolute measurement of the light levels, but is based on the local variation in the light levels on each channel R, G and B, that is to say it is not sensitive to the overall variation in power received by the sensor, due for example to the variation in the power of the source or the variation in the source/reflector distance or in the reflector/sensor distance. On the other hand, quantification of the colour videos in a small number of levels makes the device not very sensitive to the colorimetric variation of the source.
A second corner cube combination is a variant of the previous one. The optical assembly also has four adjacent corner cubes denoted Re1, Re2, Re3 and Re4, as described in
In order to identify the source reflector from the isolated reflection of a given colour, the marking differentiations described above are all used. Thus, the markings have different geometrical shapes making it possible to identify the sides of the coloured triangles of the reflectors. They are unitary or doubled and with different transmission in order to differentiate the reflectors from one another, as already described in
In the same way as in the previous case, when the source S is in the vicinity of one of the separation edges of two corner cubes, the two adjacent reflectors each generate an image on the detector, these two images therefore being partially superposed. The simpiest procedure for determining the images coming from each corner cube is to arrange a neutral filter of known attenuation on some corner cubes. For example, if the choice is made to attenuate the light levels reflected by the corner cubes Re2 and Re4 par by a factor of two relative to those of the corner cubes Re1 and Re3, a global filter of transmission (0.5)0.5 is added on the entry faces of the corner cubes Re2 and Re4.
By using coding identical to that described above, it can then be shown that, with knowledge of the codes of the surfaces of the total image by measurement, the surfaces of each of the two to four constituent images can therefore be reconstructed using their code, i.e.:
The contour of the surfaces of uncombined colours produced separately by the reflectors, and therefore the orientation of at least one of these reflectors, is deduced therefrom.
A third corner cube combination has eight adjacent solid corner cubes in the form of trirectangular tetrahedra with the same vertex O, referenced Re1 to Re8. The eight equilateral entry faces form a regular convex octahedron PRQUVW. This is represented in
Each corner cube reflector is provided with three coloured triangular filters with marking and one neutral filter, also with marking, as described above. As in the case of the previous optical combinations with four corner cubes, there are two variants of the octahedron of optical assembly.
In a first variant, all the corner cubes have the same triplet of coloured regions, which may be red, green and blue. In a second variant, the entry faces of four corner cubes have the same first triplet of coloured regions and the faces of the four other corner cubes have the same second triplet of coloured regions, which is different from the first triplet of coloured regions. By suitably distributing the colours, shapes, transmission and number of the indentations of the marks, it is possible to identify each corner cube by its image on the sensors. The solid angle covered is therefore 4π steradians, i.e. the entire space.
In a second variant, each of the eight corner cube reflectors combined to form an octahedron is provided with the same three coloured triangular filters with marking and a neutral filter, also with marking. A global filter of transmission (0.5)0.5 is applied onto the entry face of four of the corner cubes. When the marking of a corner cube comprises two indentations, the shape of the indentation closest to the vertex is always used to distinguish the side and the vertex of the coloured or white triangle of the entry face. The second indentation may have three different shapes. The images obtained on the detector thus make it possible to distinguish the reflections provided by the reflectors from one another.
A fourth corner cube combination has twelve corner cubes denoted Re1 to Re12, which are not adjacent via their lateral faces. Each corner cube reflector is provided on its front face with three coloured triangular filters with marking and one neutral filter, also with marking, as described above. The 12 reflectors with vertices O1 to O12, which are in the shape of triangular pyramids, are positioned and oriented with respect to a regular convex dodecahedron of arbitrary size, as can be seen in
The arrangement of the reflectors has the following characteristics and properties:
This
It is, of course, possible to use simple variants of this dodecahedron in order to produce less compact solutions. Here again, by suitably distributing the colours, shape, transmission and number of the indentations of the marks, it is possible to identify each corner cube by its image on the sensors.
In a variant of this fourth combination, the optical element above may be reduced to only six tetrahedral reflectors, thus covering an angular half-space, i.e. 2π steradians. The external shape of the combined reflector is circumscribed in a hemisphere,
The detection systems according to the invention are used mainly in helmet posture detection applications, the optical assembly comprising corner cubes being mounted on the said helmet. The corner cube combinations above may all be mounted on a helmet.
When the optical element comprising the corner cubes is the dodecahedron above with twelve corner cubes, one of the corner cubes is omitted, for example the reflector Re12, in order to be able to fix the optical element on the helmet. Eleven tetrahedral reflectors Re1 to Re11 are kept for the measurement.
When the optical element comprising the corner cubes is a semi-dodecahedron as above with six corner cubes, it is necessary to arrange two of them on the sides of the helmet, as can be seen in
| Number | Date | Country | Kind |
|---|---|---|---|
| 1300971 | Apr 2013 | FR | national |
