Indicator to Data Processing Interface

The technical field of the invention is that of interfacing the movement of an indicator member or stylus for manual input to data processing operations.

An interface of this type translates the location of a manually moved indicator or stylus into specific coordinates in an operating surface area. The operating surface area may be the two-dimension face plate of a cathode ray tube or other type display.

Manually moved indicator to data processing interfaces have received some attention in the art in introducing manual movements into a data processing operation and systems have been built using many physical phenomena for stylus detection. Among the more popular phenomena are acoustic, optical, force-sensing, capacitive and electrostatic. Each specific technology based on a particular phenomenon provides a different mixture of accuracy, human-factor considerations and cost to performance considerations.

The tracking of the location of a manually moved stylus by the use of a technology that is based on optical phenomena has many advantages and is receiving considerable attention in the art.

The most promising technology from the standpoint of simplicity and precision involves the reflection of light from a moving beam scanner in a plane above an operating surface over which the manually positioned indicator or stylus is moved.

The present state of systems involving optical technology requires a moving member that provides an active role in the optics and thereby introduces a limitation on the moving members. The following are two specific examples.

U.S. Pat. No. 3,898,445 describes the establishing of coordinates of a special stylus that emits a signal when a sweeping beam strikes it.

A 1971 National Aeronautics and Space Agency Technical Note describes the use of two beams located at separate points on a single line and employs reflection from a special stylus, that has a cylindrical mirror so that the retroreflected beam is confined within the scan beam and thereby permitting the shaft angle of the scan beam mounting to be usable in determining the position of the indicator member.

FIG. 1 is a simplified view of the optical system of the invention.

FIG. 2 is a graph of photodetector current with time.

FIGS. 3A, 3B and 3C are a top and two side views of one optical system of the invention.

FIG. 3D is a geometrical layout to illustrate establishing x-y coordinates for an indicator in the system of FIGS. 3A, 3B and 3C.

FIG. 4 is an illustration of a display with movement retention.

FIGS. 5A, 5B and 5C illustrate the retention, display and erasure properties of the invention.

FIGS. 6A and 6B are graphs illustrating the properties of retroreflective material.

FIG. 7 is an example detector and preamplifier circuit.

FIG. 8 is an example logic level conversion circuit.

FIGS. 9A, 9B and 9C are illustrations of analog and digital signal levels.

FIG. 10 is an example circuit for timing.

FIG. 11 is a schematic of stylus contact with a surface.

FIG. 12 is a flowchart of the data entry point selection procedure.

FIG. 13 is a flowchart of the stylus-down phase.

FIG. 14 is a flowchart of the first loss of contact phase.

FIG. 15 is a flowchart of the stylus-up phase.

FIGS. 16A, 16B and 16C are examples of the interaction of the entry point selection procedure with stylus contact variations.

FIGS. 17A and 17B illustrate the preferred optical system of the invention.

The invention employs angular information acquired through the interruption of a photodetected current produced by a scanning light beam to determine position coordinates of a manually moved indicator or stylus and employs the history of position data to select precise position coordinates. The indicator member or stylus composition is free of reflectivity limitations, the properties of different portions of the indicator may be sensed for information transfer purposes and it may even be the operator's finger.

For simplicity of explanation, the term stylus will be used for the indicator member.

Referring to FIG. 1 a plane view of the relative positioning of the elements in the optical system of the invention is provided for simplicity.

A light source 10 provides a beam which enters the system through a small hole in stationary mirror 11. The beam should be of small diameter and relatively collimated. A laser will serve as a light beam source very satisfactorily. The beam is then swept across a reference surface in the plane of the figure by a rotating or vibrating mirror scanner 12. For a portion of the rotation, the beam crosses the reference surface as illustrated by rays 1,2,3, to strike a retroreflective strip 13. An acceptable strip 13 could be made of beaded material similar to that used on highway signs or high-gain projection screens, preferably having a more directional character. Such a material reflects back toward the source a large fraction of the incident light, the amount being relatively insensitive to the angle of incidence.

The reflected light is confined to a cone having a width of a few degrees centered on the reversed direction of the incoming beams 1,2,3 as shown by the reflected multiple arrows. This light is reflected again by mirror 12 and stationary mirror 11 into collecting lens 16 which provides a focus at photodetector 17.

Under the light of continuous scanning, there will be a steady photocurrent while the beam traverses the strip 13. As the scan continues, the beam strikes stationary mirror 14 and is deflected across the reference surface from a new direction to strike retroreflective strip 15 as indicated by rays 4,5,6. The retroreflected light is again directed to photodetector 17 through mirrors 14, 12 and 11.

It has been discovered that by providing a light that is interrupted by the stylus, many limitations on stylus materials are removed and the properties of the stylus itself become available for information purposes. Even the finger of the user is satisfactory.

In accordance with the invention, a scanning light beam provides a steady photocurrent. A stylus located anywhere in the area of the reference surface will intercept the beam for some range of scanning angles, producing a dip in the photocurrent. The scanning angle at which the stylus is located is established by locating the center of the dip relative to some suitable correlation capability such as a timing mark.

Referring to FIG. 2, a graph is provided illustrating the photocurrent both with and without mirror 14. In FIG. 2, when mirror 14 is provided, the photocurrent is extended and will have a second dip when the beam is again interrupted by the stylus.

In accordance with the invention, determination of the two angles of the scanning beam where the dips appear will then permit the location of the stylus, such as by the identification of x-y position coordinates of the stylus to be established by simple triangulation.

Having established the location of the stylus, in accordance with the invention, as the stylus is manually moved, certain psychomotor limitations are encountered and a decision on which coordinates represent the most accurate data for the application involved is needed for accuracy. It has been discovered that a decision on the specific position coordinates when made employing the history of recent stylus position data provides improved accuracy.

The combination of being able to track and to specify precisely the location of an indicator member on an operating area provides the principles of a manual indicator to data processing input interface that permits a variety of applications, such as highly accurate input devices, displays, terminals and teleconferencing input stations.

In accordance with the invention, the optical system of FIG. 1 is brought into a closer relationship to the boundaries of a working area by optically folding the beams under a reference surface in the operating area. This is accomplished with 90 degree dihedral mirrors positioned along two orthogonal edges of the reference surface. A dihedral mirror is an L-shaped elongated reflecting member generally made up of two mirror strips intersecting on a line or spine at about a 90 degree angle.

A suitable optical system may be seen in connection with FIGS. 3A, 3B and 3C wherein FIG. 3A is a top view and FIGS. 3B and 3C are orthogonal side views. In FIGS. 3A, 3B and 3C, the dihedral mirrors 18 and 19 have their spines aligned with the reference surface and serve to transfer a beam from below to above the reference surface, while acting in other respects as plane mirrors. The beam that passes beneath the reference surface is shown as a dotted line whereas where the beam passes above the surface a solid line is used.

The beam from source 10 through mirror 11 enters diagonally beneath the surface, strikes the scanner 12 which is a rotating or vibrating mirror, and is deflected, still beneath the surface, to dihedral mirror 18. At mirror 18 the beam is reflected from the lower facet to the upper facet as illustrated in FIG. 3B, and thereby transfers the beam to the top where it passes parallel to the reference surface. The beam is now shown as a line over the top of the reference surface where it strikes the retroreflector 13. The retroreflection is indicated by the multiple arrows providing a high steady light for the photodetector 17.

This arrangement optically causes the single beam from source 10 to behave as if it originated from two "virtual scanners" located at imaginary locations 20 and 21. As the scan angle becomes larger, as the mirror 12 moves further around, the beam proceeds to strike dihedral mirror 19, is transferred to above the reference surface and moves along retroreflector 15. Thus, the beam behaves in these scanning angles as if originating at virtual scanner location 21.

Timing to correlate scan with angular information of a line through the stylus and a line from the virtual scanners to the axis of the mirror 12 is provided as the beam strikes phototransistors 22 and 23.

For some scan angles, the beam, after reflecting from one dihedral mirror, will strike the other dihedral mirror rather than a retroreflective strip directly. This results in the beam being brought beneath the surface again. By extending the retroreflective strips so that they are both above and below the reference surface as shown in FIGS. 3B and 3C, the beam can strike the retroreflective strips 13 and 15 at points beneath those illustrated.

It should be noted that some care is required in arranging the mirrors and retroreflectors to ensure that the photocurrent is not interrupted during the transfer from above to below the reference surface.

With the system of FIGS. 3A, 3B and 3C, almost the entire area of the surface is usable.

Where the system is to be used in front of a cathode ray tube or other display, the reference surface can be made of transparent material. It should be noted, however, that since beams pass beneath the reference surface, some physical medium such as glass is preferred to serve as a stop to prevent the stylus from inadvertently being placed in a position that will block the lower beams.

The angular information is acquired as follows.

In accordance with the invention, the folded optical system of FIGS. 3A, 3B and 3C operates to permit the beam to be considered as two virtual scanner beams originating at points 20 and 21.

Considering that the angular information desired is to be translated into x-y coordinates of the operating area, the geometrical triangulation to establish those x-y coordinates of the stylus in the operating area may be developed as shown in FIG. 3D. The stylus is assumed to be located at the intersection of two beams labelled in FIG. 3D as B.sub.1 and B.sub.2. The coordinate axes are labelled x and y and the origin point is located at the point labelled scanner which is the intersection of the incoming beam and the scanning mirror facet.

The triangulation is as follows.

Let A.sub.x be the angle between B.sub.1 and the x axis and let A.sub.y be the angle between B.sub.2 and the y axis. Let S.sub.x be the distance between the origin and the virtual scanner 21 located along the x axis, while S.sub.y is the distance to the scanner 20 located along the y axis. The x and y coordinates in accordance with the invention are then expressed in terms of S.sub.x, S.sub.y and the two angles A.sub.x and A.sub.y in accordance with equations 1 and 2. ##EQU1##

Normally, S.sub.x will be chosen equal to S.sub.y.

The angles A.sub.x and A.sub.y may in turn be expressed in terms of the scanner angle which will be referred to as A.sub.s. Assuming the scanner is rotating counterclockwise, and the beam is leaving the scanner along the x axis when A.sub.s =0. In accordance with the laws of reflection, the angles A.sub.x and A.sub.y are related to A.sub.s as set forth in equations 3 and 4.

It will be clear to one skilled in the art that the triangulation procedure will fail if B.sub.1 and B.sub.2 were to approach being parallel. This could occur at the corner of a sufficiently large area diagonally opposite from the scanner, corresponding to A.sub.x =A.sub.y =.pi./4. Near this singular point, small measuring or computing errors become magnified so that the usable area will have to be somewhat smaller than the actual available size.

As an illustration, assume that the working surface is a square of side A as shown in FIG. 3D. This leaves an unusable border of width W=(S/2)-A, where we assume S.sub.x =S.sub.y =S. Let the error in measuring angles A.sub.x and A.sub.y be of order .delta.A, and let the resulting positional errors be .delta.Q. Near the corner of the square, the errors would be expressed as shown in equation 5.

In order to assist one skilled in the art to practice the invention and achieve high resolution accuracy, the following description will be developed around some particular illustrative dimensions.

Assuming S/2=14.0 inches and A=12.0 inches. This leaves a border W=2.0 inches. Then the position errors will be about 100.delta.A (inches), with .delta.A in radians.

The optical system is subject to the following requirements. The light source must be small and collimated. The laser fills these requirements very well.

In the forward direction, a small diameter beam must scan across the tablet at a nearly constant height. The required precision for the 12 inch area example is of the order of 1 minute of arc.

In the reverse direction, the system must collect as much of the retroreflected light as possible and deliver it to the photodetector, hence the numerical lens aperture for the lens 16 should be kept large. Aiming accuracy of the lens is of reduced importance as the return light is not well collimated.

The height of the various components should be minimized in order that the apparatus be convenient to use. This may produce some conflict with the large lens aperture but an acceptable compromise is possible.

Where the light source is a laser, the following example specifications should be of assistance. A 0.5 mw He-Ne laser provides more than enough power for the 12.times.12 inch working area and is completely eye-safe. It is equipped with a beam-expanding lens and a focusing lens chosen so as to minimize the beam diameter over the surface. A diode laser is more compact and less expensive, and would probably be used in a production device.

Referring to FIG. 4, an illustration is provided of the optical system of FIGS. 3A, 3B and 3C mounted in connection with a display to provide indicator movement retention. This permits a display of the pattern that has been traced. The display may be mounted essentially coincident with the plane of the scanner so as to provide a terminal or it may be in connection with a remote location for teleconferencing purposes.

The display member shown for illustration is a cathode ray tube 25 having positioned between the operator and the face thereof, the structure of FIGS. 3A, 3B and 3C. The surface 26 is a reference surface of an operating area. It is positioned so that light from a source can pass beneath the surface 26 to be reflected by the scanner 12. The light beam is then brought to the front of the surface 25 by dihedral mirrors 18 and 19. The scanning beams strike retroreflectors 13, 15 except where interrupted by stylus 27. The system employs standard techniques in the art such that lines formed by a series of x-y coordinates of the stylus tip drawn by the pencil example stylus 27 are retained as the picture 28 is traced and erasure is accomplished by having the system recognize the different refractive equalities of the eraser end 29 of the stylus.

One example of the retention, display and erasure capabilities may be seen in connection with FIGS. 5A, 5B and 5C which, in essence, show a display screen 30 wherein a region 31 is divided into sections labelled line, enter and erase. With the capability of the invention of determining the location of the stylus, specific portions of the operating area can be assigned to communicate with the system. Audio acknowledgement is usually provided. Such portions are called "soft keys" and the audio is a simple "beep". The soft key region line is first touched and the system responds with a beep.

The system will identify the location of a point 32 as shown in FIG. 5A and display it. The soft key enter is touched. In FIG. 5B, the stylus 27 would then be positioned to identify a second point 33 while the system retained the display of point 32. In FIG. 5C, the stylus would again be positioned over the soft key signifying "enter". The soft key signifying line is touched and the system would then retain and display both points 32 and 33 and would display and retain a line between the points. If it became desirable to erase the line, movement of the stylus 27 to another soft key, assigned the function, erase will remove the line.

The system, since it operates by interrupting a beam of light, will recognize the optical properties of a stylus such as the different reflectivity of the eraser portion 29 or a difference in size or diameter of the stylus 27. Using the different optical property recognition capabilities, it is possible to achieve functions other than using soft keys.

Where the ability to recognize a difference in stylus cross-sectional size or material properties is employed, it is desirable to use a laser beam having a diameter smaller than that of the intended stylii. This simplifies the signal detection circuitry, and can be accomplished by bringing the beam to a controlled focus near the center of the surface area.

The behavior of a laser beam in the vicinity of a focus has been thoroughly studied in the art. The following is a simplified discussion which is adequate for the purpose of practicing the invention.

Assume that a laser beam of wavelength .lambda. is brought to a focus with waist diameter D.sub.o. The beam will diverge around the focus at a rate determined by a characteristic length Z.sub.o given by equation 6. ##EQU2##

At a distance Z from the focus (in either direction), the beam diameter D will be as set forth in equation 7. ##EQU3##

It will be observed that equation 6 is an even function of Z, hence the best arrangement is to produce a focus at the center of the operating surface area. The maximum diameter will then occur at the edges of the area. Let Z=Z.sub.e be the longest distance from the center of the area to an edge. With .lambda. and Z fixed, the only free parameter in equation 6 is D.sub.o. Minimizing D(Z=Z.sub.e) with respect to D.sub.o, we find the relationship expressed in equation 8. ##EQU4##

With this choice, the maximum beam diameter (at Z=Z.sub.e) will be D.sub.o .sqroot.2. In the present example 12 inch square area, the distance Z.sub.e is about 20 cm. We then find (for .lambda.=6300A) that D.sub.o should be 0.4 mm, leading to a maximum diameter of 0.6 mm at the edges. The laser and lenses must be located outside of the area, so that the distance to the focus after reflection from scanners and dihedral mirrors is about 1000 mm.

Thus, the problem reduces to forming a 0.4 mm waist at a distance of 1000 mm. A single lens would suffice if the laser divergence were 0.4 milliradians, mr. The divergence is approximately 0.8 mr, so that a beam-expander is required. This is made as follows. A lens of focal length F.sub.1 brings the beam to a focus with waist diameter D.sub.1 as expressed in equation 9.

where A.sub.o is the 0.8 mr initial divergence.

A second lens of focal length F then forms a magnified waist at the desired 1000 mm distance. The final waist location L and diameter D are given as expressed by equations 10, 11 and 12. ##EQU5##

Here, L.sub.1 is the distance from the initial waist to the second lens and L the distance from that lens to the final waist. The parameter Z.sub.1 represents the length scale for variations in beam diameter in the vicinity of the initial waist. This may be compared with equation 6.

A simple design procedure is to make an initial choice of F, the focal length of the second lens. Replace D in equations 11 and 12 with the desired waist diameter D.sub.o and L with the 1000 mm focusing distance. After some manipulation, one finds the required intermediate waist diameter D.sub.1 to be as expressed in equation 13. ##EQU6##

The focal length of the first lens may then be determined from equation 14. ##EQU7## The spacing L.sub.1 between this waist and the second lens is then obtained by equation 15.

The intermediate waist is formed a distance F.sub.1 from the first lens, hence the inter-lens spacing is F.sub.1 +L.sub.1.

It should be noted that the above equations are only valid if the lenses do not truncate the laser beam. Thus, the diameter of the first lens must exceed the input laser beam diameter. The required minimum diameter for the second lens D2 is easily calculated in accordance with equation 16. ##EQU8##

As an example, if F is chosen as 50 mm, then F.sub.1 will be 26 mm and the lenses will be spaced about 78 mm apart. The second lens diameter will have to exceed 2 mm.

These are some considerations that may be of assistance to the practice of the invention that are associated with the other components. The scanner is required to deflect the laser beam through a total angle of about 90 degrees. A polygonal mirror driven by a small motor at 60 Hz is satisfactory and is preferred because a 90.degree. angular range is considered large for most available vibrating scanners. In principle, an eight-sided polygonal scanner could provide eight 90-degree scans per revolution. However, the mirror is required to collect the returning retroreflected light. Because the facets move as well as rotate, the poorly collimated return light would be progressively vignetted or faded near the ends of the scans, leading to large variations in photocurrent. An improvement is achieved toward reducing this by using a 4-sided polygon, with oversized facets.

Returning to the embodiment of FIGS. 3A, 3B and 3C, the considerations involving the dihedral mirrors are as follows. The beam should leave the scanner parallel to the underside of the working area surface so that it will remain parallel after translation to the top surface. This will occur if the dihedral angle of the mirror is exactly 90 degrees. If the angle differs from 90 degrees by .epsilon., then the output beam angle will be in error by 2.epsilon.. If both dihedral mirrors have the same but incorrect angle, then the error can be compensated by adjusting the incident laser beam angle, or by tilting the scanner.

The dihedral mirrors may be constructed of strips of front-surfaced mirror cemented into holders machined from rectangular metal stock. The machining may be done in such a way as to ensure that the two holders would be identical, although the angles may differ from 90 degrees by a few minutes of arc which is easily compensated. The holders and other parts of the system are mounted on a flat base plate cut, for example, from jig-plate metal stock.

The retroreflectors may be made by attaching self-adhesive material to metal strips.

Several types of retroreflective material are available in the art. All are essentially semi-microscopic glass beads bonded to a metallic coating on a paper backing. The beads may be exposed, or else protected by a polymer coating. Uncoated retroreflective material has about 4 times the reflectivity of coated material, and works over a wider range of incident angles. However, it is susceptible to damage and is difficult to clean. With the system of the preferred embodiment, the signal-to-noise ratio is sufficiently large that it is possible to use protected or coated material without penalty.

The variation in reflection efficiency of the retroreflective strips with angles of incidence and divergence is shown in FIGS. 6A and 6B. There is a rapid fall-off for angles of incidence beyond about 40 degrees.

Referring again to FIGS. 3A, 3B and 3C, it will be appreciated that the light reaching strips 13 and 15 may arrive at angles less than 45 degrees. Thus, for reflection efficiency, it is convenient to machine a number of small facets into the holders for strips 13 and 15, so as to tilt the retroreflector tapes in the correct direction.

Referring next to FIG. 6B, the angular distribution of the retroreflected light, with angles measured with respect to the incident beam is shown. Although the light is reasonably directional, the cone angle is much larger than the collection aperture. The aperture is rectangular. In the horizontal direction, the aperture will be truncated either by the scanner mirror or by the gap between the dihedral mirrors. An increase in the gap size will decrease the usable surface relative to the overall dimensions. In the vertical direction, the aperture is truncated by the surface on the one hand, and by either the base plate, the scanner or the dihedral mirrors on the other. Increasing any of the dimensions will make the apparatus dimensions thicker. For the illustrative example under discussion, a horizontal aperture of about 1/2 degree is achieved, and a vertical aperture of about 1/4 degree, with an overall thickness of about 1.2 inches. The system can use the least efficient but readily available coated retroreflective tape (3M-7922) may by the 3M Corporation, in order to collect about 0.02 percent of the initial laser power.

In the practice of the invention, the x-y coordinates of the stylus location are converted electronically to data processing compatible signals through the general functions of detection and preamplification, conversion to data processing logic signal levels, timing accommodation and conversion to digital numbers.

In the detection and preamplification functions, the following example specifications are employed. The duration of a complete scan is approximately 2 milliseconds (ms), with 1 ms devoted to each virtual scanning center. During this time, a 0.3 mm diameter beam sweeps over a transverse distance of around 300 mm. The rise and fall times thus will be of order 2 micro-seconds. The photodetector and amplifier chain must therefore have moderately wide-band performance and reasonably low spectral noise densities. A satisfactory unit for these specifications is the EG&G model SD-100 photodiode, driving a low-noise one-chip operational amplifier connected in a current-feedback mode as illustrated in connection with FIG. 7.

Referring to FIG. 7, using the same reference numerals as FIGS. 3A, 3B and 3C for identical items, a small lens 16 collects the return light and focuses it onto the photodiode 17 connected between a negative voltage and ground. The photocurrent may be estimated as follows. It has been assumed that approximately 0.02 percent of the 0.5 mw laser power is returned to the detector. This is about 0.1 micro-watt. The photodetector 17 has a sensitivity of about 0.2 amps/watt at the laser wavelength, hence there will be about 0.02 microamps of signal current. At the output of the amplifier 34, the voltage signal-to-noise ratio is in excess of 100:1, although this is lowered somewhat by the fine-grained structure arising from the beaded retroreflective strips 13 and 15. Proper shielding and grounding techniques are required to avoid corrupting the signal with logic-circuit switching noise. In some cases, it may be desirable to place an interference filter in front of the photodetector to eliminate noise from ambient lighting.

Referring again to FIGS. 3A, 3B and 3C, it will be noted that the beam will be sent directly back to the detector as the scan shifts from dihedral mirror 18 to dihedral mirror 19. The amplifier 34 of FIG. 7 would be driven into saturation, requiring a long recovery time. A diode limiter is provided. Several diodes 35A-D are used in series across a large resistor 36 to provide a reasonably linear transfer function for signal currents in the normal range.

In the conversion to logic level signal function, the following considerations apply. A normal stylus is large enough to completely block the small diameter laser beam, resulting in nearly rectangular pulses with clean edges. The large signal-to-noise ratio allows the direct detection of the leading and trailing edges of each pulse without filtering. However, some efforts should be taken to eliminate direct current (DC) drifts.

Referring to FIG. 8, an illustrative circuit is provided in which the preamplified signal is fed to amplifier 37 for additional gain. The input to amplifier 37 is through a capacitor 38 which blocks slow DC drifts arising from such items as photodetector leakage currents A true zero is restored by sampling the output of amplifier 37 at a time when the scanning beam is known to be off the retroreflectors. The arrangement is similar to that used in conventional video cameras. Sampling is performed by a conventional sample-and-hold circuit 39 whose output is combined with the input to amplifier 37 with isolation by impedances 40 and 41, ensuring that zero output voltage corresponds to zero signal current. The output signal from amplifier 37 is fed to voltage comparator 42 whose output switches between logic-level 0 and logic-level 1 when a pulse is present. In the example circuit of FIG. 8, the comparison level is set manually with potentiometer 43, however, it could be obtained automatically by sampling the return signal from a retroreflector located just outside the active area.

Referring next to FIGS. 9A, 9B and 9C, the signal levels are illustrated. FIG. 9A shows the signal when a stylus is absent. The disturbance near the center of the trace occurs when the beam scans across the gap between the dihedral mirrors. The overall variation in signal level results from vignetting of the return beam by the scanning mirror and by the effect of angle-of-incidence variations on the retroreflective strips. FIG. 9B shows the signal with a stylus present. FIG. 9C shows the output of the circuit of FIG. 8 after being gated with a blanking signal.

There are certain timing functions to be maintained. Circuitry is provided to form digital numbers representing the occurrence times of the four edges, that is the leading and trailing edge of each pulse, and to pass them to a computer for final analysis. Since the stylus position is obtained by triangulation, precision is required in measuring the angles. If cost were of no concern, a precision shaft encoder could be mounted on the motor. A less expensive alternative takes advantage of low cost of circuitry in combination with the large optical lever arm provided by the scanned laser beam.

Referring again to FIGS. 3A, 3B and 3C, two phototransistors 22 and 23 are mounted at the edges of strips 15 and 13, respectively. The laser beam strikes 23 just before it begins scanning 13, and strikes 22 just after it leaves 15. These detectors 22 and 23 provide reference timing markers which have a fixed relationship to the geometry of the system. The large optical lever arm from the virtual sources allows the use of simple photodetectors and yet provides precise timing markers.

Referring to FIG. 10, an example circuit is provided. A voltage controlled oscillator, VCO, 44 is synchronized so as to provide a fixed number of clock pulses between these markers, each pulse corresponding to a precise angular increment. The VCO 44 is started by the pulse from detector 23 and its output counted by a 12 bit counter made up of 3 four-bit counters 45, 46 and 47 in series. The oscillator has a period of about 1/2 microsecond, which determines the final resolution of the measurements. The 12-bit counter generates a pulse after a count of 4096 is reached, whose time-of-occurrence is compared with that of the pulse from detector 22, a standard digital phase detector circuit 49. If they differ, a correction voltage is applied to the VCO 44 through a holding circuit 50. The arrangement is similar to a phase-locked loop, except that the error is sampled and corrected only at discrete intervals. When the loop comes to equilibrium after a few scanner revolutions, there will be exactly 4096 pulses between the pulses from detector 23 and 22. Thus, the approximately 90 degree range of the scanner will be divided into 4096 equal increments. These increments are independent of the speed of the motor and the position of the mirror with respect to the drive shaft, and nearly independent of the alignment of the incoming laser beam. Further, the oscillator 44 need have only good short-term stability. The circuit of FIG. 10 uses only inexpensive integrated circuits. The phototransistors 22 and 23 may be replaced by small vertical strips of retroreflective tape, arranged to introduce coded pulse patterns into the data stream from the main signal amplifier. These pulse patterns may be detected and used to drive the timing servo loop.

The final function is the conversion to digital numbers. A complete set of data consists of four digital numbers representing the leading and trailing edges of each pulse. These are obtained by storing the value of the 12-bit counter of FIG. 10 when the relevant edge occurs. The data values are then transferred for processing which calculates the center of each pulse before performing the geometrical computations. Referring to the signal waveforms of FIGS. 9A and 9B, it is apparent that several blanking pulses are needed to limit detection to the relevant range of angles. In addition, gating signals are needed to activate the sample-and-hold gate 39 of FIG. 8 for amplifier 37 and to initiate and terminate a `data ready` signal. These signals are obtained by decoding the binary counters. The 4-bit counter 48 is needed as some of the required pulses occur outside the measuring interval defined by the 12-bit counter. In addition, there is obtained a signal indicating successful stylus detection in which two and only two pulses are detected.

The resulting signal data is then processed in a suitable computer such as an IBM Series/1 minicomputer connected via a standard 16-bit input port. In order to further assist one skilled in the art example software instructions to execute the operations on the IBM Series 1 are contained in several subroutines. The x-y computations are provided in the Appendix sections labelled A to J. The retention, display and erase capabilities are set forth in the Appendix sections labelled K to AB.

It will be apparent to one skilled in the art that the principles set forth in the subroutines may be called by other programs where the invention is part of a larger application.

The subroutine of Appendix sections labelled A to J returns the x and y coordinates of the stylus. The device of the invention is referred to in the program as the "tablet". The subroutine consists of 3 parts that perform the following general functions.

(1) DATA ACQUISITION. This routine is highly specific to the computer hardware and to the interface circuitry. It causes the transfer of the four digital numbers representing the stylus pulse edges into main storage. The left and right edges of each pulse are then averaged to obtain an estimate of the pulse center.

(2) ANGLE CONVERSION. As discussed, the coordinates may be calculated in terms of two angles, A.sub.x and A.sub.y. These angles are proportional to representative digital numbers, N.sub.x and N.sub.y representing the pulse centers so that the angles may be expressed as shown in equations 17 and 18.

The constants C.sub.1 -C.sub.4 depend on the location of the phototransistors 22 and 23. They are obtained by a simple calibration procedure when the system is first assembled. The values are obtained as single-precision floating point numbers, stored and retrieved when the subroutine is first loaded into storage. Thereafter, the integers N.sub.x and N.sub.y are converted to floating point representation and A.sub.x, A.sub.y obtained from equations 17 and 18 above as floating point numbers.

(3) STYLUS COORDINATES. These are calculated using equations 3 and 4. The tangent functions are evaluated using a standard numerical approximation. All calculations are done using single-precision floating point arithmetic with 24-bit binary fractions. The value of S is obtained by direct measurement of the area dimensions.

Referring to the subroutines in Appendix sections labelled A to J, each routine has functional statements. The manual to data processing interface of the invention is referred to as the tablet. The regions of the operating surface that are set aside for communication functional purposes are called soft keys. If the stylus is detected in one of the soft key regions, it is recognized as not having a valid x-y coordinate. The program involves four subroutines called XY, GETXY, TAN and READXY and operate as follows: XY reads raw data from the tablet. The numbers are basically the times at which the finger was detected relative to a timing mark. The program GETXY first converts these times into angles, and performs a trigonometric calculation to computer x-y coordinates. The subroutine TAN computes tangents for this calculation. The x-y coordinates so computed have their origin located at the scanning mirror, which is outside the actual tablet. These programs are completely specific to the particular example tablet under discussion, and would have to be entirely changed if some other device were used.

READXY then takes the X-Y values from GETXY and adjusts the values so that X=0, Y=0 correspond to the lower left-hand corner of the actual tablet. It then checks to see if the values are on the main part of the tablet or on the edges which correspond to "soft-keys".

The subroutines of Appendix sections labelled K to AB illustrate the steps in providing retention, display and erase capabilities.

The example system under discussion exhibits an accuracy of about 0.01 inches when used with stylii of various diameters and tip-shapes. These include pencils, the index fingers of the user, 1-4 inch diameter discs, etc. A typical user is able to select a point with about 0.05 inch accuracy using their fingers, despite being unable to see the actual point of contact once the finger is close to the surface. This indicates the ability of the device to find the center of the large-diameter finger with precision. Greater resolution is achieved where the system, as shown in FIG. 4, is positioned in the optical path between a user and a display so that immediate visual feedback is provided.

If a stylus with a small tip is used, points can be entered without visual feedback to an accuracy of about 0.01 inches. The stylus should be held perpendicular to the surface. Alternatively, a round optical comparator with a simple cross-hair can be placed on the surface for use in entering points.

The invention provides some `human factors` considerations. The optical system senses the stylus position a small distance above the surface. As a result, it is moderately sensitive to the tilt of the stylus. This can lead to reduced accuracy if the system is operated without immediate visual feedback via a display of already entered data. When such feedback is presented, a user can easily make very precise adjustments by rocking a finger or other stylus slightly in the desired direction.

The manner of detection of the invention will sense all objects on the surface and as a result, the hand cannot be rested on the surface while operating. Any drafting aids, such as rulers or curves, have to be arranged to adhere to the surface as they cannot be held by the other hand. It is also possible to detect the orientation of a specially shaped stylus, or of two fingers placed on the operating area surface. Such information can be used in various ways, for example, to input directional as well as positional information. As discussed, the system can also determine the width or eraser end of a pencil-like stylus permitting thereby the thick end or eraser to be used for deletion.

The system, once the capabilities thus far described are available, is further adaptable to further operator-data processor interactive activity. Designs can be facilitated through the use of software-defined areas on the surface that can be touched by the stylus to direct the system to perform specific functions. Useful feedback can be provided by auditory indication, such as a "beep" device. The function direction regions of the operating area have been given the name soft keys. As previously described as an example of such use, consider entering the end points of a line. A soft key labelled `LINE` is first touched. The system responds with an audible `beep` to indicate key detection. The first end-point is touched, and is displayed immediately on a screen in a tracking mode. After the finger is lifted, the point remains displayed until a key labelled `ENTER` is touched, the system again providing a beep. The second point is similarly entered, and the system responds by drawing the line.

Points can also be entered via a pushbutton activated by the other hand.

In order to facilitate the achieving of the high accuracy capability of the invention, as the principles of the invention are translated into different apparatus, the following analysis of some typical sources of error is advanced.

One source of error is in the precision of the coordinates. A measure of the accuracy can be appreciated from the following discussion. For the example square area of side S/2, in which S.sub.x =S.sub.y =S, the equations 1 and 2 can be expressed as the following equations 19 and 20. ##EQU9##

The major source of error is the measurement of the angles A.sub.x and A.sub.y, provided that equations 19 and 20 are evaluated with sufficient numerical precision. The errors in x and y may be estimated by differentiating equations 19 and 20 with respect to the angles. Assuming .delta.A to be the magnitude of the angular errors, and .delta.X and .delta.Y to be the resulting coordinate errors. Then ##EQU10##

These diverge if A.sub.x and A.sub.y both become .pi./4, as the scan lines are then essentially parallel. Consequently, it is necessary to restrict the stylus from this region. In general, a square or rectangular surface is desirable. Assuming the operating area to be a square of side A inside a larger square of side S/2, with one corner located at the scanner as shown in FIG. 3A. This leaves an unusable border of width W=(S/2)-A. The largest errors occur at the corner nearest where A.sub.x and A.sub.y become .pi./4. At this corner, we have: ##EQU11## which when substituted into equations 21 and 22 and expanding in the small parameter 2W/S while keeping the leading terms, the errors will then be as expressed in equation 24. ##EQU12##

The position errors will be about 100 .delta.A inches per .delta.A radians.

The measurement technique described in connection with FIG. 10 provides digital measurements of each angle to one part in 2048 or 11 bits. Since the range of A.sub.x and A.sub.y is about 0.4 radians, the quantization error introduces an angular uncertainty of 0.4/2048 radians or about 0.2 milliradians, mr. Thus, random errors of about 0.01 inches are to be expected.

There is a second type of error due to motion of the origin of the light produced by use of a polygonal scanning mirror. The considerations involved in this error may be appreciated by assuming a polygonal mirror whose sides are located a distance R from the rotation axis. The deflected beam appears to originate at the point of contact of the incident beam with the facet. This point moves slightly along the direction of the incident beam as the polygon rotates. The displacement .delta.R along the beam is easily calculated in terms of the rotation angle A.sub.s. For convenience, let this angle be zero when the facet is perpendicular to the incident beam. We then find:

This motion can be resolved into motions along the X and Y axis through the following equations 26 and 27.

Thus, for the specific example under discussion, the distance R is about 0.25 inches. The maximum value of A.sub.s during a scan is then somewhat less than 20 degrees, hence the variation in X and Y will be about 0.005 inches.

In accordance with the invention, having established the accurate coordinates of the position of the stylus, it is also necessary for accurate data entry to establish the point at which the coordinate values are to be selected.

In general, manual operation may be erratic, bouncy or subject to anticipatory motion. As the precision of the positioning increases, it becomes necessary to compensate for errors introduced because of these reasons.

The solution in accordance with the invention is to use the data history to specify a particular set of coordinates at some time before the specific decision on entering occurs. This is done by maintaining a file of coordinates with the oldest being dropped as a new one is added and selecting coordinates from the file at an earlier time to allow for anticipatory motion.

The removal of the stylus from the operating area surface is an excellent data entry type signal, however, there are several practical difficulties to be overcome. First, the last few measured points are likely to be in error, the amount depending on the details of the stylus design and the skill of the user. Second, the stylus may skip across the surface when moved rapidly, producing false removal signals. Finally, anticipatory muscular motions may disturb the stylus position at the critical moment.

The procedure of the invention is based on the concept that the user will have had the stylus at the correct place slightly prior to the decision to remove it. The procedure maintains a short `history` file of coordinate values, detects liftoff with a `debouncing` accommodation and then uses the data in the history file to place the final point.

In the manual-data processing interface of the invention this is accomplished with the apparatus of FIGS. 1 to 11 which indicates the presence of a finger or other stylus, as well as the coordinate values, say X and Y, and a subroutine is provided which stores the X-Y data in a memory element called a circular stack or ring-buffer so that a previous set of coordinates can be selected. With a circular stack, or ring-buffer type of memory element, coordinate data is needed sequentially until the stack is full, whereupon new coordinate data is added at the beginning, replacing the oldest data. The effect is to maintain a record of the last N readings, where N is the stack length.

Referring to FIG. 11, there is illustrated the nature of the problem to be solved by showing a possible sequence of up/down indications as could occur as a manually moved stylus is lifted by an operator from an operating area surface to indicate a specific point. Each dot is representing one sampling point. At section A, the stylus has made initial contact, and then bounced twice. At section B, 3 bounces or missed readings occur as the stylus is moved, and then contact resumes. At section C, one more bounce occurs as the stylus is removed. It will be apparent that coordinates selected just before or just after the typical types of bounces will probably be in error. Therefore, a useful procedure should select for final processing only those coordinates measured where there is a clear contact indication such as at section D.

FIG. 12 shows a simplified flowchart of the data entry point selection procedure. There are 3 sections or phases. Referring to FIG. 12, the main portion of the algorithm or procedure consists of a loop with coordinate data being delivered via a subroutine called "READXY" described in connection with the Appendix. This loop runs until a standard sequence is detected, consisting of an initial steady `stylus-down` indication, a first loss of contact and a final steady `stylus-up` indication. In each pass through this loop, one reading is taken and then passed to one of the three section or phase subroutines for processing. The choice of subroutine is determined by the value of a variable called MODE. This variable is initialized to 1, and is then modified as appropriate by the subroutines.

The PHASE1 subroutine illustrated in FIG. 13 `looks for` the initial steady presence of a stylus. It does this by counting the number of stylus-down indications (ND), stacking the X-Y coordinates and returning to the main loop until a pre-set number, N1, is reached. N1 must be chosen long enough to fill the stack to the desired depth. If a stylus-up indication is obtained first, ND is reset to zero. This ensures that ND cannot reach N1 as the result of a series of bounces. PHASE1 terminates itself when ND reaches N1. It does this by setting MODE equal to 2 before returning to the main loop. At the same time, ND is reset to zero for later use.

The algorithm must now begin to look for a steady stylus-up indication, while continuing to stack data. The main problem is in dealing with stylus or finger bouncing, which may occur either during positioning or during liftoff. In particular, it is undesirable to stack erroneous data, which might contribute to the final plotted point if the user removes his stylus or finger soon after a few bounces. This situation is avoided by discontinuing stacking after the first stylus-up indication. This cannot be allowed to continue indefinitely, however, as the data in the stack would become out-of-date if the user moved the stylus. A constraint is set on the allowable sequence of events after the first stylus-up indication following PHASE1. In effect, the detection process is lengthened rather than compromise the accuracy of the final measurement. This portion of the algorithm is implemented by the PHASE2 and PHASE3 subroutines illustrated in FIGS. 14 and 15 respectively. The PHASE2 in FIG. 14 subroutine waits for the first stylus-up indication, while continuing to stack data. The subroutine then changes the MODE variable to 3 and sets the stylus-up counter NU to 1. Note that PHASE2 does not change the value of ND, which remains at zero.

The PHASE3 subroutine in FIG. 15 then looks for the final steady stylus-up indication by incrementing NU until a value N3 is reached.

When this occurs, the subroutine changes the MODE variable to 4 before returning. The main program of FIG. 12 will interpret this is a successful detection sequence and the loop will finally terminate. To eliminate the possibility that NU might reach N3 by a series of bounces, NU is reset to zero each time a stylus-down indication occurs.

At the same time, ND is incremented. The entire algorithm is restarted if ND exceeds a value N2, indicating either that the user has changed his mind, or that an accidental bounce occurred. The restart is accomplished by resetting NU and ND to zero, and setting MODE to 1. This prevents the data in the stack from becoming `stale` or out-of-date. Under the worst conditions, the algorithm will restart N samples after the first stylus-up indication, where N is the product of N2 and N3.

When the detection loop ends, several X-Y values, Na, are removed from the stack and averaged to yield the final coordinates. These are taken some distance back in the stack to remove any influence of the liftoff process. The optimum offset, i.e., `backup` distance from the last recorded point, No, as well as Na and the parameters N1-N3 depend on the accuracy and reliability of the stylus detection mechanism, and to some extent on the method with which a user operates the device. From a time standpoint, 0.3 seconds is a good time to accommodate the anticipatory reaction time of most users.

To further illustrate the algorithm, there are shown in FIGS. 16A, 16B and 16C several possible sequences of stylus signals. The small circles indicate sampling times. The arrows indicate the beginning of a new phase, the return to an old one and the occurrence time of the final coordinates. There is an assumption that N1=5, N2=3 and N3=3 and that there is a backup of two sampling times. Two points are selected for averaging, as shown. FIG. 16A shows an ideal sequence, while FIG. 16B shows a nearly ideal sequence with bounces only during initial placement and final liftoff. FIG. 16C shows a sequence interrupted by a bounce, leading to a restart.

The parameters N1-N3, the offset distance, No, and the number of points to be averaged, Na, have to be determined experimentally to provide good performance. There is one constraint which must be obeyed; the stack must hold at least No+Na data coordinate values before stacking is discontinued, in order to have enough data to perform the final average. However, if only this number are stored, then the first data point following the initial stylus-down indication might be included in the final average. This point is likely to be in error, as discussed previously. The possibility can be avoided by choosing N1>No+Na. It should be noted that the data point measured just before the stylus-up indication is automatically rejected, as long as No is greater than zero.

Apart from the above constraint, the parameters should be chosen as small as possible in order to achieve a rapid response time. If the device generates frequent false up and/or down indications, the parameters N1-N3 will have to be increased to provide reliable operation. Similarly, an increase in Na can improve the accuracy of a device producing large but random errors. Finally, an increase in No may be necessary for applications of the invention in which detection of the stylus is above the surface. This can occur where the finger is the stylus and the user tends to withdraw his finger slowly.

The algorithm of FIGS. 12 to 15 may be executed on an IBM Series 1 computer equipped with a storage display. In the Appendix sections AC to AG a listing of the instructions for a program capable of achieving the goals of the flowcharts of FIGS. 12-15 is provided. The set of instructions of FIGS. 18A-18E is called FINGER1 and includes some subroutine activity such as READXY described in connection with FIG. 11. The instructions also include several refinements not specifically addressed in the flowcharts of the Appendix sections labelled AC to AG. First, the computer controls a two-tone audio beeper. This provides the user with feedback at the critical steps of the detection process. A high-pitched beep is emitted when leaving PHASE1, and also when the final point is plotted, while a low-pitched beep is emitted if PHASE1 has to be restarted.

A second refinement involves another soft key region of the operating area surface, not illustrated, designated as a `sink`. Touching this area allows the user to cancel an input step entirely by moving the stylus there before liftoff. This is preferable to entering an unwanted point and then undoing the effects later via a second step.

A third refinement includes, in addition to averaging several points to obtain the final coordinates, capability to calculate the rms scatter and no action is to be taken if it is excessive, e.g., more than 3 times the normal value. This effectively rejects points taken with the stylus in motion, and discards obviously noisy measurements. A low-pitched beep is emitted for feedback if a measurement is discarded.

Although the technique of FIGS. 12 to 15 and in the Appendix sections labelled AC to AG is employed in accordance with the interface of the invention the concept can also be usefully applied in totally different circumstances. As one example, the combination of a guidance control with a triggering mechanism. As another example, the positioning of manufacturing apparatus such as a wire-bonding machine, in which it is necessary to first position a semiconductor chip by human vision and control under a bonding tool, and then to actuate the actual bonding. This, and similar operations can be improved by maintaining a file of coordinates of the chip or guidance control, and repositioning under computer control after the final manual signal but prior to performing the actual step. In this way, the actuation would not disturb the final alignment. This could be carried further by first storing all the corrected bonding coordinates as entered by the operator, and then performing all the bonds as a second step.

In the example of a weapon such as a rifle equipped with an angular-rate or angular-acceleration transducer, and a torquing mechanism, in accordance with the invention, a short sequence of 2-dimensional angular acceleration measurements would be stored until the trigger was pulled, whereupon any disturbance caused by the trigger pull would be removed by the torquer before the gun was fired electrically.

The preferred optical system of the invention is illustrated in connection with FIGS. 19A and 19B wherein dihedral mirrors are eliminated by adding a second scanner.

As the size increases, it has been found to be increasingly difficult and expensive to provide a sufficiently rigid mounting frame and adjustment to insure beam parallelism above and below the reference surface over the entire area requires considerable care.

In accordance with the preferred optical system as illustrated in connection with FIGS. 17A and 17B, a rigid beam is provided with two scanners mounted at opposite ends. Such an assembly would be mounted along one edge of an operating surface area and the remaining three sides lined with retroreflective tape. The tape does not need to be well aligned with respect to the scanners, although the scanning plane as defined by the bar must be aligned with respect to the desired operating area surface. Although two scanning mirrors are required, a single laser and/or photodetector channel can be used if the rotating mirrors are properly phased. This structure avoids a requirement for a reference surface, to prevent the stylus from interrupting any beams passing beneath the plane of detection as the beams are now confined to a single plane. This configuration is particularly well suited for use in front of a projection type large display.

Referring next to FIGS. 17A and 17B, the preferred optical system is provided. In the light of the principles set forth, it will be apparent to one skilled in the art that two similar scanners 60A and 60B are mounted on a support 61 such as a bar, a definite distance apart. The left scanner 60A is shown in FIG. 17B wherein a source 10 provides a beam which is swept by scanner 12. The operating area is surrounded by retroreflective strips similar to 13 and 15 of FIGS. 3A, 3B and 3C but covering the remaining three sides. The strips may be bowed to improve the angle of incidence of the light. The reflected light returns along the beam as shown by the dotted lines to the scanner 12 and through the split converging lens 62 and converging lens 63 to sensor 64.

A start pulse is easily obtained from the detector, as the mirror will reflect the laser light directly into the detector lens before scanning the surface. The stylus coordinates can then be obtained by a simple trigonometric calculation using the angle from each scanner and the known scanner separation distance S. Thus, if the angles measured by the scanners are A1 and A2, then the X and Y coordinates of the stylus are set forth in equations 28 and 29.

What has been described is an interface between manual operation and data processing involving the dual capability to optically determine a precise location of a manual indicator and to select which of the manually traversed exact locations represents the specific data desired. The capabilities of the invention provide both high accuracy and greater flexibility than previously available. ##SPC1## ##SPC2## ##SPC3##