GB2124850A - Rangefinder for marked targets - Google Patents

Abstract

In a rangefinder the field of view is observed from two laterally offset locations L and L' by television cameras which are used to determine two angular positions X1 and X2 of a laser spot. A narrow band filter E centered on the laser wavelength is used at one camera L' to remove scene details allowing the camera waveforms to be added giving a single observed scene in which two laser spots are present. The spot separation X1-X2 is then used to calculate range. One camera may be the virtual image L' of the other camera L seen via a mirror system M1,M2 which provides the offset D. <IMAGE>

Description

SPECIFICATION Rangefinderfor marked targets This invention relates to rangefinders and, in particular,to rangefinders in which a radiant spot is placed uponthe target to render it distinctive against the background and hence to simplifythe measurement oftarget position across the field of view seen from a base position. Such rangefinders may for example find application with robot assembly machines in a factory environment with targets known in advance.

Such a rangefinder is disclosed in an article entitled "Feature extraction ofthree-dimensional objects and fisual processing in a hand-eye system using laser tracker" by M. Ishfi and T. Nagata in the journal "Pattern Recognition" Vol.8, pages 229 to 237,1976.

The rangefinding principle employed in this rangefin deristoformthe radiant spot by projecting a pencil laser beam ata measured angleto a reference direction ontothetargetfrom a position laterally displaced from a television camera by a predetermined separation. The angular relationship between the beam reference direction atthe laser position and the optical axis ofthetelevision camera is known in advance. The position of the spot image on the camera photocathode relative to the pole of the photocathode is measured from the television camera output waveform. These facts taken with the measured laser beam angle and the predetermined lateral separation of laser and camera, enable the range of the targetspot from a base plane passing through the laser base to be calculated.An interference filter, centered on the laser wavelength, is used in front of the camera to remove all scene details from the camera output and hence simplify the measurement of spot co-ordinates.

In this method, it is not only necessary that accurate measurements be made from the signals generated by the television camera, but also that the laser beam angle be accurately measured. This means that the laser beam deflector must be accurately calibrated.

It is an object of the present invention to remove the need for accurate calibration ofthe laser deflector. It is another object of the invention to preserve scene details in the camera output so that the television camera can beusedforrangefinding andforrecogni- tion of objects within the scene in conjunction with picture processing apparatus.

The invention provides a rangefinder comprising meansforproviding a radiant spot on atargetin a field of view, meansforforming a real image of the field of view, means for measuring the position of the spotimage in the image of the field of view, and means forcalculating the range of the target spot given the measured spot image position and the parameters of the image forming means, characterised in that the image forming means comprises first and second objective lenses for forming first and second images ofthe spot in first and second image planes respectively, the optical axes of the first and second objective lenses having a predetermined lateral separation and mutual orientation, in that said measuring means determines the positions ofthe first and second spot images relativetothe poles of their respective image planes, and in that said range calculating means operates on the two measured spot image positions and the predetermined lateral separation and mutual orientation.In a preferred embodiment ofthe invention for use in applications where the image of the field of view contains scene details, the rangefinder is characterised in that the spot radiation has a spectral distribution which is narrow compared to the spectral distribution of scene details in the image of the field of view and in that a filter which passes only spot radiation is placed in the path of radiation transmitted by only one of said lenses, whereby scene details are present only in the image ofthefield ofviewformed by the other of said lenses.

In a further preferred embodiment the rangefinder is characterised in that said second objective lens and the second image plane are virtual images of the first objective lens and first image plane seen via a mirror system. The rangefinder is thereby simplified and also the measurement of both spot positions is now carried out by a single measuring means with consequent improvement in relative accuracy.

In preferred embodiments of the invention the rangefinder is characterised in that the means for measuring the positions of the spot images comprise an electronic camera photosurface in each image plane, the positions of the spot images in a raster on the photosurfaces providing the measurements of spot image positions, and means for digitising said measurements and storing them in separate storage locations in a digital computer, and in that the means for calculating the target range comprises a set of parameters defining said predetermined lateral separation said mutual orientation and the focal lengths of the objective lenses, said parameters being stored in the computer, and a stored programme in the computer for calculating the target range from the stored measurements and parameters.

It is a feature ofthe invention, when used in conjunction with a stored programme computer, that a programme may be provided for calibrating the rangefinder. The rangefinder is then characterised in that said stored parameters are derived from calibra tion measurements made on a target at a predeter mined range.

In applications ofthe invention to robot assembly machines, for example, it is of advantage if scene details can be processed and recognised as corres pondingtothestored parameters of objects which are to be brought into predetermined relative posi tions by the robot so thatfu rther assembly or measurement operations can be carried out. In particular, it is of advantage if the radiant spot can be driven onto a desired area of a known target outline in preparation fora range measurement.Embodiments ofthe inventions operating under computer control maytherefore be characterised in that picture processing means are provided for operating upon the image of the scene details to provide the co-ordinates of the outline of a target area in the raster and the co-ordinates ofthe spot image within the scene image, in that means are provided for calculating an offset between the spot image and a predetermined point inthetarget area, and inthat means are provided for actuating beam deflection means of a laser spot generator to reduce said offset to zero.

An embodiment ofthe invention will now be described with reference to the accompanying drawings, in which: Figure 1 shows a schematic optical drawing of a rangefinder with a deflectable laser for generating a radiant spot on a target, Figure 2 shows the rangefinder of Figure 1 controlled by a picture processor and computer as part of a robot assembly system, Figure 3 shows a schematic display obtained on a television monitored with signals obtained from a television pick up tube used as the electronic camera photosurface in the rangefinder of Figure 2, Figure 4 shows a schematic display obtained on a television monitor after picture processing has been applied to the signals displayed in Figure 3, and Figure 5 shows an optical schematic drawing of a rangefinderhavingtwocoplanervirtual imagesofthe objective lens.

Referring to Figure 1 a pencil-beam rfrom a Helium Neon gas laser LAS is reflected by a mirror beam deflector M to provide a radiant spot Son a target T.

Objective lens Lforms a real image ofthefield of view, including spot S, in a plane P containing an electronic camera photocathode ECP. Only the chief rays of image forming pencils of rays are shown. A partially transmissive plane mirror M1 is placed across the aperture of lens Satan angle of 45" to the optical axis 0 of lens L. fully reflecting plane mirror M2 is offset by a predetermined lateral separation D from mirror M1. The second objective lens L' is the virtual image of lens Lseen via a mirrorsystem comprising mirrors M1 and M2. Likewise, ECP' and P' are the virtual images of ECP and P respectively.In this case M2 is parallel toM1 so thatthe optical axes 0 and I of lenses Land L' respectively have parallel mutual orientatioq. The point of view of lens L' has the lateral separation D and axial separation Rfrom that of Lens L. Thusthetwo images of spot S have different separations X2 and X1 from the poles of their respective optical axes 0 and 1. The electronic camera photocathode ECP is the photocathode of a television camera pick-up tube not shown, and functions as a meansforthe digital measurement of X2 and X1 to be described later witch reference to Figure 2. Thus X2 and X1 are available as digital quantities measured from the pole of P, i.e. from the intersection of axis 0 with plane P.

The range y of the target spot S, measured as a normal distancefrom the principal plane C of lens L, may be calculated from the geometry of the principal rays in Figure 1.

If F is the focal length of lens L, R is axial separation ofthe principal planes of the first objective lens Land the second virtual objective lens L', and a and bare the lateral separations of the axes lenses Land L' respectively from the normal SAto plane C, then, by similartriangles:-

If D is the lateral separation oftheaxes of lenses L and L', D=a-b from which,

The two system parameters FD and R may be determined from calibration measurements made on a target at a known distance Y as follows. The laser beam r is deflected to provide a set of pointsT,T' etc on a screen placed across the field of view in a plane B parallel to plane C and at a known range Yfrom it. A set of valuers X2, X1; X2', X1,, etc are thereby obtained.

Expressing equation I as a linear relationship between two variables X1 and X2; YX2 - (Y + R)X1 = FD, this being the equation ofthe line of bestfit passing through the points X2,X1; X2',X1' etc.

from which the system parameters FD and R may be found as:

Thus the system parameters are determined in this calibration procedure after the apparatus has been set up and need not be known in advance. When the apparatus is set up initially, the mirrors M1 and M2 are made parallel by observing a pointtarget at a distance very large compared to D and adjusting the mirrors until the two images of the target are coincident in plane P. The adjustment is checked for a number of very distant pointtargets distributed across the field of view. The co-ordinates of the pole ofthe photocathode are obtained by noting the position of the image of a distance source placed on the optical axis 0. The pole provides the datum forX1 and X2 measurement.

If lens L is of adjustable focal length, a so-called zoom lens, the calibration procedure can be repeated throughoutthe range of focal lengths available. Pairs ofvalues of FD and R arethen obtained and can be stored for use with any desired setting of lens focal length. Longerfocal lengths can be used to obtain improved range accuracy with a given value of D and a given precision of distance measurement in plane P blithe electronic camera photocathode. Shorter focal íerigthswould be used to obtain range measurements in widerfields of view.

From equation I, it should be noted that if R = 0 i.e. if the principal planes ofthetwo lenses are coplanar, then the range equation simplifies to:

The range now depends on the difference ofthe two image spot measurements and no datum forX1 and X2 measurement is required. Such an arrangement may be provided by creating anothervirtual image L" of L, coplanar with L' using a second pair of plane mirrors as shown in Figure 5.The range y is measured from the coplanar principal planes ofthe virtual' lenses L2 and 4. In Figure5 L1 is the virtual image of L in the partiallytransmissive plane mirror M1 and L2 is the virtual image of L1 in plane mirror M2. L3 isthe virtual image of L in a plane mirror My set at 45 to the axis 0 of L. A plane mirror M4, parallel toM3, is offset from M3, oppositelyto M2, by a separation chosen to placeL4,thevirtual image of L3 in M4,coplanarwith L2. D is now the lateral separation between virtual lenses L2and 4.

The use of a single lens and measuring system has the obvious advantage of requiring less apparatus.

Additionallythe errors in the measuring system are common to both cameras, reducing overall system errors especially in the case of the apparatus of Figure 5.

In Figure 1 a filter E is placed to intercept radiation entering lens Lvia mirror M2. Conveniently filter E is placed between mirrors M1 and M2to reducethe diameter ofthefilter needed to coverthe aperture of lens L.ThefilterE is chosen to have a narrowspectral passband centred on the laser wavelength. The image ofthefield of view formed by lens Lfrom radiation entering by way of mirror M2 is thus substantially devoid of scene details and contains onlythe image of the spot S.The image in plane P thuscomprisesascene imagewith a spot image on the target together with a second spot image displaced from the first spot image and on a side of it determined only bywhether M2 is displaced on one side orthe other of M1. Thus the second spot image can always be identified as being a predetermined one ofthe pair of laterally displaced spot images.

As will be described later, the positions of the spot images relative to the pole ofthe image plane can be measured in spite ofthe presence of scene details in one ofthe scene images. However, the radiant spot on the target may be provided by other means than a narrow laser beam. In a factorywork area, the target may, for example, be a gripper used to pick up and move components, for example, as part of an assembly robot. A light emitting diode (LED) may be attached to the gripper in known relationship to the component when in the gripper. The LED can be energisedto provide the radiantspotwhen a range measurement is needed and, in this case, general scene illumination provided by controlled lamps may be removed leaving onlythetwo spot images in the image plane.Alternatively, a spot of fluorescent phosphor may be provided on the target, energised on demand byflood illumination with ultra-violet radiation. If a laser is used to provide the spot, an additional mechanically moveable filter of narrow pass band may be provided to intercept all radiation entering lens L,thereby removing all scene details when range measurement is required.

The rangefinder is especially suited to relatively short ranges as might befound in afactoryworkarea.

Ranges of the order often to a hundred times the objective lens lateral separation are probably the best magnitude of ranges which can be measured with useful precision using a television camera as the spot image position measuring means. Howeverthe ranges which can be measured and the accuracy obtainable depends on the resolution and precision ofthe measuring means. Mechanical measuring means may be employed using for example, a pair of laterally spaced photodiodes to determine the centre of a spot image. The photodiode difference signal is zero if the spot image is split equally either side of the photodiode separation line. The difference signal provides an error signal which may be fed to a servomechanism carrying the photodiode pair, thereby maintaining the photodiode pair locked to the spot image.A second similar servomechanism may be locked onto the second spot image. The relative positions of the two servomechanisms may be measured by high resolution optical means, for example, a diffraction grating measuring system.

The rangefinder of Figure 1 provides the range y of the target as the normal distance to the principal plane C ofthe lens L. If the slant rangeto the centre of the lens aperture is required, a further calculation is required as follows:-

Referring to Figure 2 the rangefinder is shown with all its system parameters under the control of a programmed computer COMAs an an example, the rangefinder is shown as part of an automatic assembly system in which a rectangular block componentT in the jaws of a gripper G is to be inserted into a recess H in another component T1. As part of the process of aligning T over H, the ranges of TandT1 are to be determined.

The field of view containing T and T1 may be illuminated on demand by lightsources LSfedfrom power supplies PS underthe control of COM. The laser LAS may be energized by power supply LPS under control of COM. The laser beam r is deflected by a pivoted mirror M driven by a deflector DEFto an angular position supplied by COM. The electronic camera photocathode ECP is the photo cathode of a television camera pick-up tube PU, for example a "NEWVICON" (Trade Mark) tube. The scan coils SC are energised by a scanning generator SG to provide a standard television raster on ECP. SG also contains means for supplying to COM the instantaneous line number ofthe raster and the digitized value of the instantaneous position ofthe raster spot along that line.

The video signal output of PU is binarized in a threshold unitTHR in which the binarizing threshold level is underthe control of COM. The binarized video output is supplied to a picture processor PP controlled by COM and connected to supplytheco-ordinates of selected picture details in the raster to COM.

suitable picture processor PP may comprise the pattern edge finding apparatus described in British Patent Specification 1,401,008 which is adapted to find the edges of binarized patterns.

The computerCOM includes a processor PRO, a volatile read-write memory RAM fortemporarily holding data for processing by PRO, and a read-only memory ROM for holding programme information for conditioning the operation of PRO. The bus BS connects the processor PRO to the read-write memory RAM, to the read-only memory ROM, to the lamp power supplies PS, to the laser beam deflector DEF, to the laser power supply LPS, to the scan generator SG, to the threshold unitTHR and to the picture processor PP.

First, it will be assumed that the laser spot has been energized and deflected onto a desired partofa target, say T, by means to be described later, and that the range of this target is required. The first part ofthe programme in ROM sets a firstthreshold value in THR and may, optionally, switch offthe lamps LS to remove scene details. The binarized picture is then processed in PP to find the co-ordinates of any outline present.The dimensions of a binarized laser spot image are known in advance and are stored in ROM and it is also known that the laser spot will be the brightest object in the scene. If first threshold value produces an outline having dimensions larger than the known spot dimensions, the computer COM is conditioned to increase the threshold value fed to THR and to re-examine the outline dimensions.

When, by successive approximations, the outline is reduced to two separate areas of the known spot dimensions, a programme in ROM is accessed for calculating the co-ordinates X1 and X2 ofthe centres ofthese areas in the raster. The next part of the programme is for COM to access the system parameters FD and R stored in RAM and obtained from a previous calibration measurement. The values of X1, X2, FD and Rare then entered into a programme stored in ROM for calculating the range according to equation land outputing the rangeto RAN for subsequent use in the automatic assembly system.

It is a useful feature ofthe rangefinderwhen used under the control of a picture processor and compu terthat desired targets may be identified by the geometry of their binarized outline by comparison with stored values and that the laserspot may be driven onto a desired place in the target and ranges found entirely automatically.

Figures3 and 4 illustratethe nature ofthe operations carried out by the picture processor. In Figure 3 the unprocessed television picture is shown.

The gripper G and blockT are shown with the laser spot S centrally placed on T. Also shown are the images G' andT' ofthe gripper and block as they would have appeared butforthe presence ofthefilter E. The secondvisible image of the spot S' is shown. In Figure 4 after picture processing, onlythe blockT and the spots S andS' are visible, the gripper and any other picture details having been eliminated as having outlines which are not stored in the computer COM.

Also in Figure 3 the componentT,with its recess H is shown. The spots shown are those whichwould be seen during a subsequent rangefinding operation on T1. The spot S forT1 is shown locatedoffthe centre of T1 and the second spot is shown closerto Sthan for theblockTindicatingthatT1 isfurtherawayfromthe camera than T. The picture processing operation determines the outline ofT1 and enables italso to be selected as corresponding to a stored outline. The location of S within the T1 outline is also determined and a displacement calculated which would bring the spot to the centre of T1, an area previously selected as being accurately related in depth position to the vertical walls ofthe recess H.Using the currently available measure of slant range, target bearing and picture displacement, an angular deflection is calculated for DEFwhich will bring the spot to the center of T1.After the deflection has been made, the spot position in relation to the outline centre is again calculated by COM from the new picture processor output and any correction of spot position effected thereafter.

Thus,the picture processing and the rangefinding provide the co-ordinates ofthe block Tin relation to the hole H so that displacements may be calculated forthe gripper G which will place blockTwithin hole H.

It is also a useful feature of the rangefinderwhen used under the control of a picture processor and computer that the system parameters given by equations II and III may be determined in an automatic self-calibration routine using a target screen at a known range.

Claims (17)

1. Arangefindercomprising meansforproviding a radiant spot on a target in a field of view, means for forming a real image of the field ofview, means of measuring the position of the spat image in the image ofthefield ofview, and meansforcalculating the range ofthetarget spot given the measured spot image position and the parameters ofthe image forming means, characterised in thatthe image forming means comprising first and second objective lensesforforming first and second images ofthe spot in first and second image planes respectively, the optical axes of the first and second objective lenses having a predetermined lateral separation and mutual orientation, inthatsaid measuring means determines the positions ofthefirst and second spot images relative to the poles of their respective image planes, and in that said range calculating means operates on the two measured spot image positions and the predetermined lateral separation and mutual orientation.

2. A rangefinder as claimed in Claim 1, characterised in that the spot radiation has a spectral distribution which is narrow compared to the spectral distribution of scene details in the image of the field of view and in that a filter which passes only spot radiation is placed in the path of radiation transmitted by only one of said lenses, wherebyscene details are present only in the image ofthe field of view formed by the other of said lenses.

3. A rangefinderas claimed in Claim 1 or Claim 2, characterised in that said second objective lens and the second image plane are virtual images ofthe first objective lens and first image plane seen via a mirror system.

4 A rangefinder as claimed in Claim 3, characterised in that the mirror system comprises a semitransparent mirror across the aperture ofthe first objective lens on the side ofthe lens remote from the first image plane and inclinedatan angle to the optical axis ofthe lens and a mirror offsetfrom said first lens optical axis by said predetermined lateral separation and inclined to provide the mutual orientation of said optical axes.

5. A rangefinder as claimed in Claim 3 or Claim 4, characterised in that said first and second objective lenses and said measuring means are coplanar virtual images of one objective lens and one measuring means respectively.

6. A rangefinder as claimed in any one of the preceding Claims, characterised in that the means for measuring the positions of the spot images comprise an electronic camera photosurface in each image plane,the positions ofthe spot images in a raster on the photosurfaces providing the measurements of spot image positions and means for digitising said measurements and storing them in separate storage locations in a digital computer, and in that the means for calculating the target range comprises a set of parameters defining said predetermined lateral separation, said mutual orientation and the focal lengths ofthe objective lenses, said parameters being stored in the computer, and a stored programme in thecomputerforcalculating the target range from the stored measurement and parameters.

7. A rangefinder as claimed in Claim 6, characterised in that said electronic camera photosurface comprises a television camera pick-up tube.

8. A rangefinder as claimed in Claim 6, characterised in that said electronic camera photosurface comprises a solid state array of photodetectors scanned to provide said raster.

9. A rangefinder as claimed in Claim 8, characterised in that said array comprises a rectangular array of photodetectors arranged in rows and columns regularly spaced to cover the real image area.

10. A rangefinder as claimed in any one of Claims 6,7,8,9 and 10, characterised in that said stored parameters are derived from calibration measure ments made on a target at a predetermined range.

11. A rangefinderas claimed in any one ofthe preceding claims, characterised in that said means for providing the radiant spot on the target comprises a laser providing a narrow parallel beam of radiant energy and beam deflection meanswherebythe beam can be directed onto a target in the field of view.

12. Arangefinderasclaimed in Claim 11,char- acterised in that picture processing means are providedforoperating upon the image ofthe scene details to provide the co-ordinates of the outline of a target area in the raster and the co-ordinates ofthe spot image within the scene image, in that means are provided for calculating an offset between the spot image and a predetermined point in the target area, and in that means are provided for actuating said beam deflection means to reduce said offsetto zero.

13. Arangefinderasclaimed in any one of Claims 1 to 10 inclusive characterised in that said radiant spot comprises a self-luminous source on the target.

14. A rangefinder as claimed in Claim 13, char- acterised in that said source is energisable on demand for rangefinding.

15. A rangefinder as claimed in Claim 13, characterised in that said source is a fluorescent spot on the target energised on demand by incident radiation.

16. A rangefinder substantially as described with reference to Figures 1 and 2 ofthe accompanying drawings.

17. A rangefinder substantially as described with reference to Figure 5 ofthe accompanying drawings.