CA2091395A1 - Method of and apparatus for touch-input computer and related display employing touch force location external to the display - Google Patents

Abstract

ABSTRACT

A method of and apparatus for determination of touch location on a display screen or the like or other surface embodying a force-sensing platform or surface supporting or otherwise externally contacting the display screen monitor apparatus, and responding to the forces created by the thrust of touching a point of the display screen, to sense and calculate the location of the touching point. The underlying technique employs force-sensing means responsive to all six degrees of freedom of applied (touching) force and torque, achieving force location away from the plane of the sensors and in spite of tangential force components by calculating the point of least magnitude of the three-dimensional torque vector from among all points within the screen or surface, and outputting this point as an estimate of the intersection point of the screen or surface with the thrust line of the touching or other contact force.

Description

METHOD OF AND APP~RATUS FOR
TOUCH-INPUT COMPUTER AND REL~TED DISPLAY
E~PLOYI~G TOUCr.~ FORCE LOC~TION
E~TERNAL TO THE DISP~Y

The present invention relates to touch screen cathode-ray tube and similar displavs, as for use in computers and other video svstems and the like, being more particularly directed to novel methods of and apDaratuS for determining the touch force location on the display from apparatus disposed exterrlal ~o and re.~ote from the dis~lay, as distinguished from ~orce sensors applied to the periphery and/or over or adjacent the display surface itself.
I~ore generally, still, the invention relates to novel three-dimensional orce locating techniques adapted fcr ~easurement of forces applied outside the plane or surfaces of force-sensing elements.

aackground of Invention Though thus more general in application, one of the important uses of the invention is in the field of computer or related display screen systems, such as cathode-ray tube displays (or LCD, LED, electroluminescent or other 3 ~ ~

electro-optical displays or the like); and it is therefore to this exemplary use t~at the invention will hereinafter be described as an important illustration.
A modern computer typically presents its user with such a display screen on which may be presented descriptions or pictoral representations of various choices or selections which the user may make. In many cases, the quickest, ;- easiest, and most intuitive way for the user to respond is by physically finger-touching the areas of the screen which show the desired selections.
To allow this, the computer must be equipped with an input device which permits the program on which it is operating to determine the fact and location of such touch events. For presant purposes, any input device or this sort will be ter~ed a "touch screen".
; A desirable touch screen input device should be inexpensive, rugged, reliable, and sufficiently accurate.
It is also very desirable that a single model work with a wide range of different display devices, and that it be susceptible to easy field installation by untrained users, either on new or on e~isting equipment.
Unfortunatelv, e~stlng touch scre~ns, such as ehose ~ .
:, .

.
.~

~ 3 later described, are or relatively low manufacture volume and thus very expensive by the standards of their natural market, being therefor2 precluded in major usage from integration at the time a display is manufactured. Tn addition, they require great effort, expense, and manufacturing expertise to retrofit. Since each model is more-or-less unique to a specific screen geometry, differeint models must be made in great profusion, or would-be users must be restricted in their display choice. For a combination of functional and cosmetic reasons, thus, certain prior art touch screens are indeed built into the display device, such as a cathode-ray tube, at initial manufacture (though expensive, due to low volume), and others require an awkward retrofit (also ex?ensive). Such prior touch screens, moreover, are closely tied to the design of the display device with which they are to be used, and must be provided in a profusion of different types to find wide application. Many, furthermore, have inherently expensive sensor structures tightly constrained by the geometry, compatibility, and packaging constraints of the associated display, so that sensor structures cannot often be ootimized for cost.

,};

.
"
' .: : ; ! ~ ~, .

Turning to such prior art techniques for determining touch location on a cathode-ray tube or similar display screen, they lnvolve some combination o distributing sensors around the peripher~ of, or over the surface of, the actual displaying surface or screen. Such known methods employing force sensing to locate the point at which a force is applied to a surface generally embody three or more Eorce sensors placed in a plane, but not allowed to lie along a single line. The axîs of sensitivity of each is oriented perpandicular to this plane, and the outputs of the sensors are used to compute the location of contact forces which are applied in this sarne plane. If and when the contacted surface is allowed to depart from this plane, the unpredictable tangential components of the contact force must necessarily cause errors in the reported location. If the contact surface lies far from the plane of the sensors (or is severely non-planar), prior methods are inef'ective.
Specifically, a first system of this nature is adapted for the front portion of cathode-ray tube screen displays, being provided with various additions to enable touch localization, including Doth resistive and capacitive sensing technologies, 1n which an s.Ytra ssnsor plate lS

' `:
,, .

. .. ~; : .

s applied over the face o the display screen. The plate bears one or two layers o transparent conductor patterns which deveLop and convey touch location information to conductors at the edge of the overlay plate. While efforts are made to keep all components trans?arent, losses in practice are sufficient substantially to reduce image brightness and clarity. Examples of such touch screen sensors may ~e found in U.S. Patents Nos. ~,l38,539;
4,293,734; 4,353,552; 4,371,746; 4,806,709; and 4,821,029.
A second approach involves surface acoustic wave (SA~) technology in which a glass overlay plate carries acoustic energy generated, redirected, and sensed by transducer and refiector means disposed about the periphery. Touching the plate damps this energy in a manner particular to the contact location, as described, for example, in Eleograp'nics 1981 flier "Surface Acoustic Wave".
Another technique has involved a planar force sensing technology in which piezoelectric force transducers support a glass overlay plate, attaching it to a mounting.
The intersection of a finger-touch thrust line with the -transducer plane occurs at a point which is associatea with a specific ratio of transducer outputs, allo~ing tùe , ."'i,;

:. .

: , . .
':" ~ : . , . . :

. ,, position of this point within the plane to be computed.
When curved, phosphor-bearing screen surfaces must necessarily deviate from the plane, creating a particular form of parallax error in which the user, expecting response at a particular point, instead actually receives response at another point. Sensor techniques and signal processing suitable for such an approach are described r for e~cample, in U.S. Patents Nos. 4,340,771; 4,355,202 (and prior art strain gauge sensors described therein including U.S. Patent 3,657,475 and "One-Point Touch Input of Vector InEormation for Computer Displays," C. Herot et al., Computer Graphics, Vol, 12, ~lo. 3, pp. 210-216); and 4,675,569.
Still another approach uses planar force-sensing technology in which steel beam springs with strain gauge transducers constitute force sensors bearing the entire weight of, for example, the cathode-ray tube assembly. This technology avoids the image degradation of an overlay plate, but at the cost of requiring greater sensor dynamic range and problems of rejection of stray signal,s from swav and vibration. Its function is otherwise substantially identical to the above-desc-ibed piezoelectric system. U.S.

Patents ~1Os. 4,918, 62 and 5,038,142 describe such a sysi:em, "

, ` ' . ~ :": ~ :

o~ ~

citing, also, earlier piezoelectrlc and related sensors.
- Infrared light tschnolo~y has also been proposed in ~hich many separate beams travelling from emitters to detectors define a plane. When the user's finger (or otAer probe of sufficient width) crosses this planer the identity of interrupted beams locates the "touch". Again, a transverse component to the touch motion can lead to a parallax error in which response at the expected location is replaced by response at an unexpected location. Parallax errors Eor this technology tend to be particularl~ severe, since the response surface cannot be positioned to intersect the phosphor surface, nor be shaped to conform to it.
~dditionally, such apparatus may require obtrusive bezels.
; An example of such a systsm is described in pages 12-~1 of text entitled "Caroll Touch", which also summarizes the before-described resistive-ca2acitive sensor overlay ~ systems, surface acoustic wave svstems and piezoelectric . systems, as well.
Each of the above methods has an effetive response surface which, unfortunately, fails to be coincident with the active surface of the display, leading to the universal p ri or pe r f ormance impe r f ec t Lon o f pa ra l la x .

I

~' ,,, , . . ,: : , :

3 ~ ~

The before-described resistance, capacitance and acoustic plate sensors have a response surface which conforms to the actual physical surface of touch contact, such lying visually about 1/2 inch in front of the phosphor surface in the case of a cathode-ray tube display. An operator whose eye is somewhat to the side, will therefore perceive an error in the touch svstem response unless touching a surface point that lies directly over the desired target point, rather than the target point itself.
The piezoelectric and o~her planar orce-sensing systems above-described, on the other hand, do not actually report an actual location of surface contact, but rather provide what may be called an "indicated point" on a "vi~tual response surface". The indicated point is at the intersection of the thrus~ line and the plane of the orce sensors. For the described infrared beam system, such an indicated point is where the finger breaks the plane of the infrared beams. Since the glowing phos~hors are not located in such plane, the virtual surface does not correspond to anything visible or intuitive, making the parallax error of these devices particularly troublesome.
Underlying the present invention, however, is the .
. , .

" ~

:
, _ discovery of a novel method of and apparatus for enabling a r~ide variety of cathode-cav tube or other sc.een display systems, as in computers, monitors and other video systems and the like, to be placed upon or in touch with a common, universal force-sensing platfor.~, the sensors of which are thus exter~al to the plane of the display screen and remote even from the display equipment itself, but nonetheless provide a novel three-dimensional force locating techniaue ~or forces, such as the finger-touching o~ the display screen, while obviating all o the above-described limitations and disadvantages of the prior art techniques, including the total elimination of parallax.
Other distinguishing features of the invention from the above-described and other prior art approaches ~ill be more fully addressed hereinafter.

Obiects of Invention A principal object of the present invention, âccordingly, is to provide such a new and improved method of and apparatus for touch screen sensing, void of the limitations of prior art syst~s, and, ~o the contrary, adapted ~or unootrusive location o~ the sensing external ~o ' , ~ .: . ,, the display, preferably beneath or ln back of it, and universally employable ~it~ a wide variety OL displav systems of many different configurations and tvpes.
A further object is to provide such a novel touch-locating input device for use in conjunction with a computer display, to locate touches directed at features of the displayed image; and which, in addition to unobt~usive location external to the display, can easily be field installed, with one or a very few types or sizes adaptable to all displays. Such design, furthermore, which can be optimized for low cost, since unconstrained by internal design aspects of the display, is robust, long-lived, and immune to wear, providing parallax-free response for any display surface, and ~ithout degrading the dis~layed image.
Additionally, it is a further and more seneral o~ject of the ~nvention to provide a novel method of fully locating the thrust line of a force in three-dimensional spac~, or the line of minimum torque, accurately determining the location of a force applied to a surface, or the location at ~nich a force is dir-cted through a surface. Such surface may be far removed from t;~e plane of the sensors, mav ~e substantially different from a flat plane sur_ace, and is , .

;~, :

,~

.

not constrained by device design to have a par~icular relationship in space to the device.
In accordance with tne invention, moreover, a device embodying the same may be programmed or calibrated in use to project a virtual response surface of any shape to anv location, subject only to certain natural limitations. Such limitations are that the thrust lines of the forces to be localized shall intersect the response surface with positive polarity at but a single point (or rnore precisely that the lines of minimum torque magnitude do so), and that the object bearing the physical surfaces to be rnatched by the virtual projection, shall be appropriatelv coupled to or supported by the measuring device, with the distances and forces involved falling within the dynamic range and sensitivity of the particular measuring device.
Other and further objects will be explained hereinafter and are more particularl~ delineated n the appended claims.

Summar~ of Invention In summary, however, in one of its important ao~lications, the invention embraces a ~ethod o~
detertination oE t~ucn locetion on a disp ay sur~ace :
. ~

: .
"

, . . - ~ .
,. , , , ;
, , . ~ ~: . : .: .

apparatus, that comprises, contacting the display surface apparatus against or in touching relationship to a force-sensing platform having sufficient degrees of freedom and sensing sensitivity to develop and encode, in response to the thrust of touching a point of the display sur~ace, the coordinate components of the resulting thrust vector and the components of the accompanying torque vector;
calculating rom the encoding, a location on a line of .;
minimum torque to provide an intersection of the display surface; and outputting the resultiny sensed location as an estimate of said touching point.
In the `oest mode and preferred form of the invention, a six-degree force platform or plate is employed upon which any monitor or other display device may be placed, receiving line power and sending a single parallel or serial port cable to the host computing device -- say, for example, an T~M ~C (personal computer) or the like. The format may be something like an electronic bathroom ~Jeight scale, but reading out si~ numbers at once instead of one. These encode the same information as is contained in the ~, v, and z coordinate components OL thrust, and the roll, pitc~, and ~iaw components of ~orque. For convenience, the actua1 .

i .

': ' ' ' '~ ' "'',`'' ' ":, ' ".'`' '1"' :~ .
., ;. ,, ; , .
,'' ; ' . ' ', ''~' .: ' ~' ' 3 ~ ~

numbers are an equivalent linear transformation of these.
The challenge in recovering a touch position 'rom such a remote sansor platform or surface lies in the fact that the direction of touch force on the display screen can vary greatly from one instance to the next, even when exactly the same point on the screen is touched. When the sensors cannot be confined to the same effective plane as the touch ~as is done with, and lndeed required by prior art devices, as before explained), different touches at the same point of the screen may produce different sets of numbers.
The present invention admirably obviates these problams, however, by ta!cing two important considerations into account, the appreciation and application of which are at the heart of the invention.
First, the force at the point of contact can be described quite accurately as a pure thrust. For present ., purposes, the torque components referenced to this point are negligible -- partly because the area of contact Ls small, and in part, because the finger is not attached to the screen. This kind of force is referred to 'nerein as a "simple con~act force'', defining the "thrust line" as the locus of points obtained by extending the hrust vector .~ .

' ^
.
, :' 1' ' ' .. ~' ~ , : , through the point of contact. For a simple contact force then, the line in space of points ~ith minimum (in this case, zero) magnitude of the three dimensional torque vector is coincident with the thrust line of the force.
Secondl~, a measurement of the thrust and torque occasioned by the touch at some remote reference point is sufficient to reconstruct the line of minimum torque, and therefore the thrust line. (For simplicity, in the discussion that follows, the term "thrust linel' is sometimes used to refer to the line of minimum torque which approximates it. The method oE the invention, however, deals directly with the latter.) Although the theory and oractice of this will be more fully developed below, consider first a brief outline of the principle involved.
The thrust (or perhaps more intuitively, the reaction thrust to maintain static equilibrium) is an invariant of position, but the torque is not. The torque vector is perpendicular to the plane containing the thrust line and the reference point, and has a magnitude equal to the product of the thrust masnitude times the distance at closest approach of the thrust line to the reference point.
Since the directions and magnitudes oE the thrust and torque .
:~;
, L~ 9 S

vectors are obtained by measurement, one can, in summar~, calculate backwards as ollows~ ind the direction perpendicular to the plane containing the thrust and torque vectors ~which direction of t~o is determined by consistent use of some handedness rula); (2) Proceed in this direction a distance equal to the magnitude of the tor~ue vector divided by the magnitude of the thrust vector, ending up at the point on the thrust line which is closest to the reference point; (3) Extend the (known) thrust vector through this point to obtain the thrust line which, of course, intersects the surface o the display screen in a single point. The contour of this surface either is known, or is conveyed to the computsr through an appropriate calibration procedure enabling the location of the touch .

point.

Other details o best mode design and construction are ` more fully dascribed hereinater.
., ~

Drawings The invention will now be described in connection with the accompanying drawings, ~ig. 1 of which is a side elevaLiGn depicting the use or the Eorce-s~nsing platfor~. of .

.
:
'~ ~

.: - " ~, ; ,, 3~

the invention as a remote touch screen system for a computer or similar monitor with a catnode-ray tube dis?iay screen supported on the platform;
Fig. 2 is a top elevation of the platform sensor of Fig. 1 depicting a simple means for reproducibly locating the supported monitor upon the force sensing platform of the embodiment of Fig. l;
; Fig. 3 is an isometric view of the major components and the construction of the force sensing platform, showing the same in open position;
Figs. 4 and 5 illustrate a design for the springs used in ths platform;
Fig. 6 is a cross-sectional view of the details of a suitable pair of capacitive displacement sensors for the ~lat~orm;
Fig. 7 is a simplifled schematic circuit diagram of '~ electronic conversion and calculating circuitry for the system; and Figs. 8-10 provide graphical depictions Oc the force vectors and geometry of the force locating operation for a force of the kind made locatable by the inventlon, and which is applled out of the plane of the platform sensols.

' .

i , 3 '~ ~

Description of Preferred_Embodiment(s) A siY degree-of-freedom ~orce sensing platform 32, Fig.
1, is used to provide information sufficient for the calculation of an "Effective Thrust Line" resulting from a "Simple Contact Force'' arising where the monitor display screen 31 (or other object, in general) supported upon the platform, is touched or contacted by another object, as by the finger F. The platform also contains electronic signal conversion and calculating means suitable to prepare and d01iver desired output results to external devices, as over, Eor example, a simple RS-232 serial communication link 38.

Force Locating Platform Construction Fig. 1 depicts the platorm embodiment 32 of the remote force-locating device constructed in accordance with a preferred form of the present invention, contacting or touching, indeed supporting, the ~ase surface of the cathoda-ray tube monitor 31 on support surface 33 to provide a touch screen function, though remote from the cathode-ray tube screen itself which is touched by the user. The force locating device 32 receives powe~ thYouah an AC adapter cable 35, and communicates location informa~ion to a :, . .

`:

3 ~ ~

computer (typically a personal computer, or "PC", not shown) through, for e~ample, the cable 38.
Since the calibration of the locating function depends upon the position of monitor 31 with respect to the platform 32, the platform is provided with a stop 39 (see also Fig.

2), which is an "L" shaped strap or land of material protruding above the platform surface. The monitor 31 is - slid back and to the right against the stop 39, giving an accurately and rapidly reproducible position. Dashed-circle sets 40 and 41 illustrate two possible patterns of foot location for two possible monitor styles.
; In Fig. 3, the platform 32 is shown separated or opened into an upper plata assembly 50 and a lower plate assembly 51. When brought together and fastened with screws 53 and ` washers 54 (onLy one set shown), a flange 55 overlaps flange 56, so that the four steeL beam springs 52 in the corners carry the entire weight of the upper assembly and all supported objects. Only under conditions of overload, do the flanges contact each other or the opposing plate, so as to protect the beam springs 52 ~nd nsreinafter d scussed ; caDacitor sensors 57 (having upper and lower segments 57a and 57b) from damage. The u~per capacitor elements 57a face , - . . ~ .' . :

3 ~ ~

and allgn within the lower capacitor elements 57b to provide linearly independent measurements responsive to all six components of plate-to-plate displacement. These capacitance sensors are shown provided substantially midway along the ront of, and toward the rear of the two sides of the platfor~ plates. Six wires 58a provide connection of the upper elements to printed circuit board 60 through 7 connector halves 58b and 58c. Similarl~, 59a, 59b, and 59c provide connection to the lower elements. Connector 61 provides power, allowing the printed circuit board electronics (not shown) to compute force location data which is then output~ed through connector 62.
The beam spring 52 is shown enlarged in Fig. 5. It may i be produced from a double-L flat 70 of Fig. 4, folded, as shown, and provided with press-it threaded inserts 71.
Fig. 6 shows details of a pair of suitable and pre'erred capacitor displacement sensors 57 in section. The plates l~ay be Eormed from r~ctangles of unetched printed circuit board material, for example, about 3 square inches in ~rea. Foil capacitor ~lates 72 are supported on - insulating laminates 73, which in turn are attached by adhesive to an upper bracket 74 and a lower bracket 75. The : . . . .

. ~ , :~ :

brackets 75 and 74 are nesting brackets, which are shown flattened parallel to the platform plates 50 and Sl to which they are respectively secured. The free arms of the brackets are bent outwardly (~or 14) and correspondingly inwardly (fc,r 75) to mount the pair of capacitor plates 72, oriented at matching angles (sho~n as 45) to the platform.
Capacitor 76a, formed of the right-nand capacitor plates 72, Fig. 6, is sensitive to relative capacitor plate displacements along axis 76b orthogonal to the capacitor plates; while capacitor 77a, formed of the left-hand capacitor plates 72, is sensitive along orthogonal axis 77b.
The two sensitive axes themselves are thus at r.ght angles.
'~
Operating Circuit Explanation Fig. 7 provides a simplified schematic diagram of the electronic conversion and calculating means incorporated in a successf~lly operated platform force sensing device 32.
~licroprocessor system 80 may be one of many different standard designs, such as the Intel Type 80188 with associated components, physically comprising one to several integrated circuits, and logically comprising a processing unit, read/write memory, firmware program memory, a small ;, : . , , ",~

..
. ;
' ':: ' : . ~ :

21 2~

non-volatile read/~rite memory for storage of calibration and operatlng mode data, an asynchronous serial I/O
capa~ility for driving output caDie 38, a digital input capability for receiving the output of analog-to-digital (A/D) converter 81, and a digital outp~t capability for setting the input selection of a multiplexer 82.
Timing circuit 83 divides a 20 MHz clock by 128 to give . .
156 ~Hz 5V square wave signal 84 for sensor drive, and by 65,53b to drive converter 81 to provide one 16-bit convers.ion every 3.3 ms.
Signal 84 is connected to each of six idential capacito- impedance measuring circuits 85. An operational amplifier 86 generates a signal 87, which transfers charge through the sensor capacitor 57 exactly equal and opposite to the charge flowing through fixed capacitor 88, thus maintaining virtual ground at its summing junction 89. The peak-to-peak amplitude of signal 87 i5 thus linearly proportional to the capacitor plate separation of sensor 57.
A resistor 90 of high value ;22 MOhm, for e.~ample) provides a return path for input leakage, kee?ing signal 87 ~it~in the operating range of amplifie- 80. The value of capaci~or 88 (5 p~, -or e~ample) is chosen approxi~atel~r to match the , : ' ' . " . :, , ', -~ ' S~ 3 ~ '~

value shown by the sensor capacitors under conditions of no platform load. A synchronous amplitude detection circuit 91 converts AC signal 87 to DC signal 92, which, in turn, is admitted through multiplexer 82 to A/D converter ~1 when processor 80 selects this channel for measurement.
Connections 94 lead to other circ~lits similar to 85. A
complete set of measurements across all si:c inputs may be completed about fifty times each second.
To achieve the desired accuracy, the force platform 32 must be able to measure contact forces of a few ounces to a relative accuracy of about 1~. These must be measured in the presence of a large but unpredictable static load: i.e., monitors commonly weigh as little as twenty pounds to as much as eighty pounds or more. It is necessary, therefore, to find a design in which wide load range does not comoromise either economy or the necessary sensitivity.
~ Since the smallest practical gap for capacitor sensor ; 57 is about 20~ of the no-load opening, t~e amplitude of signal 87 may vary from about SV P-? (peak-to-peak) down to about lV P-P at the ma~Yim~m monitor welght of 100 lbs.
Circuits 91 and a2 being unitv 3ain, the working in~ut ranse ~ of converter a1 is ro~ghlv I to 5 VDC. wit~ approprla~e :, ...

, ~ , . . . -operating margins allowed, tnis provides a sensitivity of about 500 counts/lb. Sincs the ~S noise is a~out 1 count, force changes of 3 ounces and larger can be resolved with 1 or better accuracy, based on a single conversion value difference from baseline. As the typical touch force determination is based upon a weighted average of a number of measurements, the actual minimum orce is somewhat less?
Single-slope A/D converters may be of known design which can combine nigh resolution with low cost. Converter 81, as indicated, may have 16-bit resoLution; but as a counterpart of its very inexpensive design, it has nowhere near the linearitv or freedom from drift requirsd for 16 bit absolute accuracy. Its non-linearity, however, is considerably less than 1~, and its worst case drift is less than a count per minute. As the firmware in processor 80 re-calculates baseline every few seconds or less, drift is thus obviated as a source of er-or. Since, moreover, it is the relative error of small c~anges that is of concern, not absolute error, the linearlty is entirely adequate.

~nalysis The cesired remote three-dimensional Lorce locating ~ .: : .
., ,:
:, : : :.
.: : .. . :;

3 ~ ~ ~

2~

(rinser-touching) function is achieved by the above exemplary embodiment in the following manner.
In one mode, data specifying the Effective Thrust Line itself (or, more particularly, the effective line of minimum torque) may be the desired output.
In another, the intersection of the Effective Thrust Line with a known surface contour may be computed, and the coordinates of this point within the surface outputted, perhaps in combination with other detected charactoristics, such as force magnitude. The known surface contour may match the actual physical exterior of the supported monitor or other object, or of a portion of it, in which case the coordinates identiy the actual point of physical contact.
The force-sensing platform 32, in its most basic form, thus comprises two rigid plates 50,51 supported and separated by at least three springs 52 placed around the peripher~. These springs, as above indicated, are preferably securely fastened at both ends so as to prevent all pivoting or sliding motions which might introduce mechanical hvsterisis through friction. Thev are designed to o~fer (when so mounted) roughly ecual spring rates in both compression and shear; such rates tvpicaliy being in .: , , , , ~ - . . .. .
.. :. ~ - : ,.. : - , :
- ; ~
' ! ' ,' . :
.. ' ' ' ..
,i.
', : ' ~ ', ' ,' .
', . ' . ' 6~ 3 ~ ~

the range of a few one-thousandths of an inch per pound for each spring and in each direction. The overall spring rate is chosen as a trade-ofE between the greater sansitivities which can be achieved with a softer mountins, and the greater freedom from dynamical errors achieved when a stiffer mounting raises the resonant frequencies of the supported mass.
The capacitance dis~lacement sansors 57 are mounted between the plates with positions and directions of sensitivity chosen effectively to encode all six degrees of freedom of rigid motion. These sensors, having the prefera~ly variable seometry achie~led through their previously described construction are inexpensive and sensitive. In another embodiment, each of the six sensors may comprise an optical emitter-detector pair mounted to one of the platform plates, the beam of each being variably attenuated by a small piece of yraded transparency film supported from the opposing plate (not shown).
, ; The sensor outpu~s are detected, scaled, and muLtiplexed to l-orm the input to the A/D converter of Fiy.
7. ~his, as earlier stated, may be of very inexpensive single-slope design wnile still providing the requirad wiae : , , - - , . : :
. . ~, . .
; . , ~

- : ::
.. . .. :~ . ~ ' : . . :

dynamic range, since neither high stability nor perfec.
linearity is required. Digitized values sensitive to the various displacements are fed to tne standard microprocessor system 80, which performs the necessary calculations and formats output as required by the application. In the case or this touch screen application, this may include emulation of other touch screen devices, as well.
:~. Thus, the touching of a point P on the display sc~een 31 of Fig. 8, will result in a thrust mechanically conveyed to the remote force-sensing platrorm 32, that, through the ; six degrees of freedom and sensing sensitivity thereof, will sense and develop encoding signals corresponding to the x, and z coordinate components of the resulting thrust vector, . Fig. 9, and the accompanying torque vector roll, pitch and yaw components. As shown in Fig. 10, as hereinafter more fully explained, the before-described microprocessor calculations will derive the remotely sensed location of the touching point, and output this at 38.

Theory of Use of Force Data While Fig. 8 depicts the display device 31 resting upon the force sensing platform 32, Figs. 9 and 10 re-represent ~, .

.

:: . ! , , . . ".

:: . : ' ' "
': ~ ` :'": ' ' :
':, : . , ~ ` ' ,'.' :, 3 ~ ~

this same system, but with the display and platform replaced by a transparent cube for diagrammatic clarity. For concreteness, a specific reference poin~ "R" is shown, with a particular coordinate systsm illustrated at 104 in Fig. 9.
Thrust and translation vectors may consist of an x, y, z enumera~ion of pound or inch values, and torque and rotation vectors may consist of a ~itch, yaw, roll enumeration of pound-inch or radian values. Although centering and aligning the coordinates on the axes of symmetry in the manner suggested by Fig. 9 would make numerical examples of the matrices discussed below look simpler, this choice is otherwise arbitrary. For simplicity, moreover, force and sensor outputs will be discussed as though only time-varying components existed, since carrying through such constants as the display weight or the baseline sensor outputs would unnecessarily clutter the description without altering the results.
In Fig. 10, thus, a thrust vector THRUST_P i5 diagrammed at its point of application P. The "thrust line"
of the force applied at P is defined to be the locus of polnts reached by the infinite eYtension of THRUST_P through touch point P. The act~al area of contact between the 28 ~ 6~ ~ ~

user's finger and the display screen actually consists of many points close to P, through which infinitesimal contributions to the total thrust pass in directions roughLy p~rallel to T~RU~T_P. This means that the torque e~erted by finger pressure about P, and indeed about all points on the thrust line, is negligible. Since the torque magnitude of the force referenced to other points rises in proportion to..
their distance from the thrust line, there exists a well defined line of minimum torque magnitude which is virtually coincident with the thrust line for a force of this kind --called a "simple contact orce". (Note that were the hand, instead of using 'inger ~ouch, inserting a peg into a pegboard, the contact force might not be simple; the mechanical interlock of peg in hole would allow a ..
substantial uncontrolled torque to be transmitted through ; the "point" of contact. Indeed, if the peg and hole were square, there would not need to be any relationship ~t all between the thrust line and the line of minimum torque).
As will be seen, the force measurements made by the platform are sufficient to computa the line of minimum ; torque. The eYternal surface oE the display, however, ls also required to remair. in fi~ed reLationship ~o the forse ~:~
:~

.,. :. - , "' ;. :.

2~39~3 -~ plate since the last performance of a user calibration procedure. (This procedure is, in effect, a way of letting the platform know where the sc.een is). The sensor data, therefore, is logically sufficient to locate the contact point of a simple contact force in three dimensional space, and, given appropriate calibration data, any two dimensional grid imagined on the surface.
Returning to Fig. 10, the reference point R has been selected to express the aggragate effect of the time-varying forces on the sys~em. For greater clarity, the plane containing thrust line 102 and reference point R is made visible by rectangular segment 105, wit~ sides parallel or perpendicular to the thrust line, and by the intersection 106, where this plane passes through the boundary of the cube representing the display and force platform.
The particular total force, later discussed, referenced as "TF P@R", comprises THRUST R and TORQUE_R taken together, and which, applied at point R in Fig. 10, would 2roduce the same motions and displacements of the top plate of the ~latform as does the touch force at P. It is a known result, in fact, that there i5 always a unique equi~alent total orce oE this kind tor ~ny reierenc~ point chosen.

.:, ... . . ..
.. . . . . .

: . : . : ~ .: :: . :

3 ~

rOr present purposes, it is convenient to imagine R located at the center of s,v~metry of the suspension syst~m inside the force sensing platform. ~Since this is a point in empty space, one must imagine it connected to a massless rigid extension o~ the top plate).
j The force at R equivalent to that at P is expressed by the relations:
T'dRUST R = T~RUST_P (la.
TORQUE R = TORQUE_P ~ R->P cross l'HRUST_P, ( lb.
where "R->P" is the dispLacement vector from point R to point P, and "cross" refers to the vector cross product.
Sincs, for a simple contact forcQ, TORQUE P is effectively zero, TORQUE R is perpendicular to plane 105 and has a magnitude given by the produce o the magnitude of T~RUST_P
times the length or vector R->Q. (Q is found by dropping a perpendicular 107 to the thrust line). Consider the ' following equation for the location vector "R->T" of a point "T":
R- >T = Lambda * THRUST R
~ (THRUST R c~oss TORQUE R) / ¦THRUST R¦~2, (2.
: where paired vertical bars are understood to return the ~ magnitude ot tne aector between them, and vnere the symDols ;' , ,.
.
~ ~ .

: , . . . ..
.

"*", "/", and " " represent t~e appropriate forms of multiplication, division, and exponentiation, respectivelv, and wher2 "Lambda" represents a s~alar parameter. The cross product in the second term on the right constructs a vector in the direction of R->Q, with magnitude of R->Q * THRUST_R * TORQUE_R , such that the whole second term can be seen to locate the point Q with respect to the reference. Since the first term represents an arbitrary length vector in the direction of THRUST_R, whicn is also the direction of THRUST P, T takes on the identity of each and every point on the thrust line for some value of Lambda.
In the more general case, lt can be shown that the eauation or T generates the line of minimum torque; but given the constraint that TORQUE P be zero, this is indeed the same as -- ~
the thrust line.
It has thus been shown that the point of contact can be calculated from information sufficient to deter~ine the total force vector acting on the system as seen at some reference point, such as R. Let us now turn to the relationship between this total force vector and the values measured by the plattorm sensors.

. ~ . .

i: l !~
:,l 3 ~ ~

Aquisition o~ Force Data The thrust and torque on the system produce a displacement of tne top plate whi.ch may be expressed as a combination of a rotation about R, followed by a translation. The three component rotation vectcr is represented as "ROTATION R", and the three component translation vector as "T~ANSLATION_R". The total disolacement vector "D_R" is also defined as consisting of the components oE translation followed by the components of rotation.
In the range where Hooke's law applies, the deflection is desc~ibed by fl~xure matrix "FLE~MAT R":
D_R = FLEXMAT_R * ~_R_MEASURED, (3.
; where "TF R MEASURED'I is the sum of all forces, referred to R, except for the non-baseline spring forces. It is distlnguished from TF P@R in recognition of the non-equilibrium effects to be discussed in the next section.
Consider one particular sensor located at a point "S", the response of which is characterized by a sensitivity vector "SENSITIVITY_S". ~1hen the rigid extension of the top plate at S moves in the direction of SENSITIVITY S, the sensor gives a maximum positive response which is equal to ;
' .,.. - ; ' , ~ , :

.

:~`
2~9 L3~

the product of the distance moved times the magnitude of SENSITIVITY_S, When the motion is perpendicular to this line, there is no response; that is:
Response_S = SENSITIVITY_S dot TRANSLATION_S, (4.
where "Response S" is that one of the six components of the sensor data vector "~ESPONSE" due to the sensor at S. (The operator "dot" is the vector dot product). In the limit or small rotations, the geometry of the system gives:
TRANSL~TION_S = TRANSLATION R
+ ROTATION_R cross R->S (5.
The error is about one-'nalf the rotation magnitude, in radians, times the result. Since the rotations oE interest are less than one thousandth of a radian, the error is insignificant compared to desired accuracy. Taken together, the previous two relationships imply that the response is a linear transformation of the total displacement, the de~endence being summarized in a 6 by 6 matri.Y "SLNSL~1AT R":
RES~ONSE = SENSMAT_R * D R, (6.
The if, by definition, a 6 by 6 calibration matrix "CALL~AT R" is given by:
CALMAT R = inverse (SLNSMAT R * FLE.YMAT_R), (7.
there results:

, , .. ..
, ~ .

~, ~

2 ~ ~ ~3 ~ 5 TF_R_MEASURED = CAL-~AT R * RESPONSE. (8.
That CALMAT_R be tractable requires that both FLEYMAT_R
and SENSMAT_R be reasonably far .rom singular. For FLEXMAT_R, this means that the s?rings should have roughly comparable compliance in both compression and shear. The~
also should be spread apart a distance something like the size of the touch surface, to gi.ve a reasonable balance between torsional and translational stiffness. For SENS~AT_R, it means that sensors should be placed and oriented to respond as independently as possible. Again, how spread apart tAey are determines the relationship of ro~ational to tr~nslational sensitivities, the desirable balance being set by the touch surface size.

Interfarence from Intertial Effects To this point, it has been assumed that forces are applied slowly and smoothly enou~h closely to approximate static equilibrium. In reality, the non-zero compliance of the display and platform imply a difference between 'ITF R ~EASURED'I, the actual force sensed by the platform, and TF_R, the total force mathematically projected from point P. This diEEerence may be represent-d as:

;

. ~

3 9 ~ ::
i : :

; 35 TF_R ME~SURED = TF_P@R f TF_R_INERTIAL. ~9.
"TF_R_INERTI~L" is the reaction orce of the display and top plate mass referred to R. It consists of excitations of the normal modes of vibration of this mass.
It has power spectrum confined almost entirely to frequencies above a value somewhat below the lowest nor~al mode frequency.
. .. .
It would be desirable to use standard linear filtering techniques to remove the corrupting influence of TF R IN~XTIAL. These techniques comprise taking various time-~eighted averages of the measured data. It must first be demonst-ated, however, that such avarages will not disrupt the accuracy of contact locaLization in some other wa~.
Consider the time evolutlon of a typical touch force.
It not only rises and falls, but constantly changes direction. As the fan shape swe~t out by the instantaneous thrust line will usually have some conical cupping to it, the thrust line of a summary average forc~ does not .ecessarily lie close to any of the instantaneous values.
Given that P itself does not move, however, (R->~ in equation lb is constant), it can be seen that a time .
~,, .
i.,i 1 :

.~ .

~,: ' . ', ' '~.', ~ "' ' . ':; ' '.', '.'.' æ~3~

weighted average of the total force components at R, or of any linear transformation of those COmDOnents~ corresponds to the components (or translormed components) of a similarlv time~weighted average of the instantaneous forces at P. But any sum of forces applied at P totals to a force at P~ 50 the thrust line computed from the time weighted component.s ("efLective thrust line") must ~ass through P.
Now a linear filter applied to TF R MEASURED will produce a response which is the sum of TF_P@R filtered and TF R INERTIAL filtered. The latter is close to zero for an appropriate filter, and, as above shown, the first term provides values which compute to the correct contact location, thus yielding the desired result.
An effective filter may be of known lowpass and/or notch design, preferabLy i~plemented digitally within the microprocessor system 80. Such a filter can have a grou~
delay as low as 0.5 to 1 times the cycle time of the lowest normal mGde of vibr~tion, or something in the range of 0.1 second. As this is shorter than the typical touch duration, good measurement amplitude is maintained (i.e., the power s~ectrum of the touch lies in substantial part in frequencies lower than those of TF_R INERTIAL), and " , .~ .

., ': ' ' . . , : , easonable r~sponse s?eed s achieved. .~ote ln pa~tic~lar ~hat this ~rou? cslav is c~~en ~uch short~r tnan tne dam3ins time of the system -- th~ clted vlbrat-ons ~a~ ring for many c~fclss befor~ somethl~n~ aporoximating static e~uillbrium is ac:~ieved.

The Planar i~odel There are many situations where a contact surface ma~
oe adequatel~/ apDro~imat~ y a proper7 ~ cated 'lat ?lane.
It is found that good results may be achieved in applyirlg this soecial case to a touch apolication, if the ma.~imum deviation of t.~e sur ace ~om the plan2 does not exceed abou~ 3 times tne re~u rec, a-_u~c~; i.e., for most touches, the tangential component OL the contact Lorcs is one-third or less of the normal one. ~lthough ?ractical difficuities in the placement o, sensors within this same ?lane may elevate th2 cost ~nd limit .he a?olicabllity of t;~e befor2-desc-lbed prior art techn~'cues in many such aDDlications, the method of ths invention for calculating an electrical . ^ ~ ~ ~. ~ .. : ~ ~ ~ â ~ a 2 n, a ~ _ O C ~ ~ ~
c~ ~s ,,o .r~ ~ e a ~ ~ 'f a _ ~ _ _ '' _ ?~ ? ~ ~ ^ ~-.; âOi ' _ :C..... .
;' .

'i~ '' ~ . .
, ., : , ~ ' .. ` , : i ' , , 3~

Consider t;~at t.~e contact surface is to be labeled bv a t~o-dimensional ~rid ~it:~ coordi!tates "u" and "~J". The orrgi-~-o~-t~is- gri-d~-is at-~o~i~t !'O~ n.- three-dimen~ionaL
sDace, with whic~. we associate the three dimensional basis vectors ~U and EV. I tne point o contact "P" is at coordinates <u,v~ within the grid, we may writs:
R->P = P~->O ~ u*EU f v*EV. (10.
.,~
Now it can be shown t~at there e.Yist three sets of 5i~
numbers, represented bv the Si.Y cornponent vectors U_CAL, V_CAL, and W CAL, such that (in the static limit):
u = (U_CAL dot RESPONSE) / (r~_CAL dot RESPONSE) (lld.
v = (V_C~L dot RESPONSE) / (W_CAL dot RESPONSE) (llb.
and "W_CAL dot RESPONSE" is proportional to the normal component of the contact force. For brevity, define:
i us = U_CAL cot ~ESPONSE (12a.
vs = V_CAL dot RESPONSE (12b.
= r~ CAL dot RESPONSE. (12c.
As "us", "vs", and "w" ultimately are just linear trans~ormations oE TE~ EASURED, the filtering d~scribed ; aDove ~a~ be a2?L_ed _ ~es2 ce~-Jed data st~eams. Tnen tb.e eauations:
u = (filtered us) , (f ltered ws) ;i3a.
,~

. , . . .. . ,-,~ . . -; . , . , , , . ~ . .. . . . . .
., ~: . .. .
, - . . : , , ~
,: ~ - : . , ::

3~ ~

v = (filt2rsd vs) / (Liltered ws) (13b.
closel~ a?proximate u and ~ without requiring static equilikrium. The "filt2rQd w'' may be monitored to determine the presence of contact; and well defined values of u and v may be calculated from the above equations wh2rever "filtered w" is large enougn.
Consider now how values for ~_C.~L, V_C~L, and W_CAL can be obtained. After placlng the display device in its position on the force-sensing platform, the user runs software which takes the user through a cali~ration procedure. This soft~are ~ay run on the host computer, if desired, rather than on mic.oDrocessor system 80. After the procedure is completed, the calibration values are downloaded via communic~tion link 38 for storage in a small non-volatile memory which is part of ~0. The system lS then ready for use.
For convenience, let it ~e assu~ed that the grid which is to be used on the display screen has coordinates <u,v> =
<0,0> at the lower left corner, and <u.~> - <1,1> at the uo?er -igrt. lhe caiibration -~s21f _ar be p2-^o--ed as follows: The four ?oints at ~ne four corners of -he screQn, <0,0>, <0,1> <1;0> and <1,1>, ar2 successivelv illu~inat2c, ., . ~

.

, ~ ~ : , ' ,, ' ., , ~

3 9 ~

and the user is ins,ructed to press each one, tnrQe seDarat~
times as it a~pears. The user may be further instructQd to del~ver touches with an intentional and varying direction of sidëwavs forcé, as this arrows ~o~rAmore accurate~-calIbration~
of the response to tangential components. The exact force and direction of each touch is not important, `nowever; onlv - that each is placed carefully at the indicated point.
For each of the six measurements made ~ith u = 0, it must be that "U_CAL dot RESPO~S~" is also 0, since "w" is certainly not infinity. Thus U CAL is a vector in the null space of the matrix made by collecting together these si:~
measurements, and a scalar multiple of this can be extracted by standard methods, called "U_CAL_A". A similar multiple of V_CAL, "V_CAL_B", can be determined from '~e touches with v = 0. While any arbitrary multiple o~ the calibration vectors taken together suffices, the relative scallng must be consistent. Define:
a = U_C.~L_A / UCAL (14a.
b = V_CAL_B / VCAL. (14b.
Di~idlng lla by llb, then ~ultir~lvin~ both sides bv a/b, we gQt ror the touches at <1,1>:
a/b = (U_CAL_~ dot RESPONSE) / (V_CAL_ a dot ~ESPONSE) .
;''` ' .
:

- ., ~ , .

' `
~ .' ' ~ , ' .'~

The value of a~b is det2rmined from one such touch, or as the averase of the ratios so derived. Then, V_CAL A = (a/b) * V C.~L 3 (16._ _ Using ~quation lia, for éach oi- the si~Y touches~wïtn u = I~
Z_CAL .A dot RESPONS2 = U C.~L_A dot RESPONSE. tl7.
Collecting together the six numbers computed from the six right hand sides, and pre-multiplying this vector by the inverse of the matrix of the corresponding measurements, Z CAL A is extracted and the orocess completed.

A Non-Planar ~odel Now to consider briefly ar. aDproach to the more general non-~lanar case.
At the factory, each platform can be pre-calibrated in a specially designed fixture which supplies a set of six precisely known forces. The forces are chosen such that the matrix of these forces, each ex?ressed ln terms of a speciLic refarence point and coordinate syst~m, such as 2 and 104, is readily invertible. The matrix OL measurements is t.hen muiti~lied 'v t:-is inve~se, vieldins the desir2d cali~ration matrix (C.~L.~'~T ~, aoove), Ir~nic.î is stored in th2 non-volatile memory.

.. :

~ .. . . . . . .
.

3 ~ ~

In the field, the user calibration procedure presants the user with a point at the center o the displa~, and each of the four points centered along one edqe. Two differently directed touches are reauested for each point, and the point locations in space determined from the points of intersection o the thrust line pairs. Since tne line will not precisely intersect, the mid-point of the segment is used which is perpendicular to both as a surrogate "intersection''. If the segment is too long, or the lines of the pair are too close to parallel, the user will be prompted to repeat the point. That plane, vertically oriented cylinder, and sphere are now det2rmined which best fit ~in the ~S sense) the 5 test points. The quality or fit for each is compared and the shape passing closest to all the points is retained for use. (These three families tried here are by far the predominant geometries for display surfaces).
In application, the factory calibration matri~ may be used to compute the thrust line in accordance with equation 2, aDoYe, wl~h the use ol appropr-ate .11t2r1ns o~ ~he sensor data. ~he inrormatlon from the user cali~ration is then emploved 'or calcllatlng t.~e point of surface , . ~
. . . . , , ~
.

: . . ~ ., - . : . ~

.. . . . . ..
. . . .,,: :. . . .
:: . . , , :
, . : .:
-.

.3~

; intersection in three space, which is re?orted via 38 ln terms of the two-dimensional coordinates of that rectangular plane grid, which when orthogonally projected onto t;~e postula-ted surfacs,"places the 5 ~est points in the right place.
To recapitulate, in the application just described, . e~plicit use is made ~ichin the embodiment itself of such entities as the thrust line, and the com?onents of the total force at R, which were developed in the analysis. In the application to the planar case, however, they need not appear within the embodiment, although they were used to develop it. Thus, it may be seen that two different ty~es of embodiments within the scope of the invention may employ calculations that may differ radically as to both structure '~~ and detail. '~hat they do have in c~mmon, in accordance with the invention, is:
(1) Use of force-sensing means responsive to all si~
degrees of rigid motion; and (2) Calculating means which from the output of sald 0rc~-s2nsing means, computes the 13cation of a contact I f3rce; such computed lccation being substantiallv fr~e of error caused by the ?resence of an unpredictable t~ngential .`

9 ~ ~:

component of the contact ~orce, ~or all potential contact points of interest, including those wei:L removed from the plane of the sensors.
Thë information provi.ded bv the proposed out-of-plane-sensors of the inventlon is in fact theoretically sufficlent to eliminate errors from the `oefore-described tangential ~orce component. r~hlle particular practical techniques for performing each stage of ths required calculations have been ~resented, it is to be understood, nowever, that there are many difEerent ways in which these calculations may be performed, and many variations in such matters as the location zr.d orientation of sensors, ty~e or sensor, type of support, etc.

, . . .
Recapitulation of Distinguishment rrom ~rior Art In summar~y, thus, there are at least three major ways in which the methodology underlying the present invention distinsuishes it from the previously described and other prior art techniques and whic~ are responsible for the novel results attained W''Lh the inventlon.
~ irs~, the inven~ion emplo~s ~orce-sensin~ means responsive tc, all si:~ desrees of f-sedom or applied force .... ~ ........................................... ..

.

.

.. : ~ -2~139~

and torque. Prior art methods, on the other hand, go out OL
their way to be sure that they are not responsive to tangential components.

.. .. ~... . .
Sacondly, the present inventnion achieves force location away from the plane of the sensors, in spite of such tangential components.
Thirdly, the invention computes the point of least magnltude of the three-dimensional torque vector from among all points within the surface of interest, and then out2uts this point as an estimate of the intersection point of the surfacs of interest with the thrust line oE a contact force.
That this is, for all dis~ositions of this surface, distinct from prior art methods or sensing or calculating, is further explained below.
Each of the above characteristics clearly distinguishes the present invention from the before-described prior art and the results obtainable thereby.
Prior art methods, before expLained, on the other hand, report the posi~ion within the flat plane of the sensors at which the magnitude of a _ertaln two d mensienal tor~ue vector is zero. This vector may be vie~ed as the projection onto the plane of the sensors, at each point in space lvlng ' .
-- , :, : : .: : " : :

., :~., :, ' '` ' 3 ~ ~

on that pLane, of the true three dimensional torque vector at that point. Not only is the method of the invention far more general, in allowing the surface of interest even to be curved, and indeed remote-from the pIane of the sensors, but it is also conceptuall~ and numerically distinct from prior art when a~plied to a flat plane which may contain the sensors.
To clarify this, consider the following: support a flat board at its corners with sensors operated in accordance with prior art. Place this whole apparatus in turn upon a devi.ce of the invention, so that locations o contact upon the board may simultaneousl~ be read out by both m~thods. Drive a screw into the board at 45 degrees to the surface (or at any angle that is not perfectly perpendicular). Again, press the screwdriver, also a~ 45 degrees, against the screw head, but without twisting. At this point, both ~ethods will report the correct contact location.
Both two- and three-dimensional torque vectors ~re zero at the Qoint of contact. The 'ield of the three dimensional tor~ue vector can be visualized as c~lindera of equal lensth arrows centered on the thrust line, tne arrow length ~or ;

a7 each cylinder rising in proportion to the cvlinder's radius.
The individual arrows lie per~endicular to the thrust line and pursue each other around it in ~ ci~cular pattern.
Now, as the screwdriver is twisted to the right, a non-zero torque vector aDpears at the point of contact which points into the board parallel to the thrust line. This component appears uniformlv added throughout the ~~ield, lengthening the arrows everywhere and bending them to point somewhat in tne direction OL the thrus~ (they r.ow appear to pursue each other in right-handed helicesl.
Since the minimum magnitude still lies along the thrust line, where only the parallel component is pres2nt, the method o~ the invention alone continu2s t3 -eport the correct contact point. The two-dimensional projection of the parallel component at the point of contact cannot be zero, since it is inclined to the surface normal (as it must always be, to at least some extent, for any real force).
Away from the thrust line, however, the helical inclination of the torque field causes the two-dimensional projection of some particular vector to vanish at an e~traneous point.
Imagine the board horizontal and the screwdriver inclined towqrd the user, ~ith a line dra~n on tie ~oard ~' ......

2~39 L39~
~ 48 .~
through the point of contact, extending to the right at rignt angles to the thrust line. Pick a point on this line, ---say-,-2,-i.nches.~.fr,o~.,..the..sc~ew...,Press~on ,the scr~w with a 1 pound force, while applying 2 pound inches clockwise torque wi~h the screwdriver. The torque component occasioned by the thrust has magnitude 2 pound inches and points Lnto the board inclined ~5 degrees toward the user. The toraue component occasioned by the twisting has magnitude 2 pound inches and points into the board incllned 45 degrees awav from the user. The resultant has magnitude 2.818 pound ; inches, directly normal to the board. Its projection into the surface, therefore, has zero magnitude; t~is is the , location of the extraneous polnt re~orted as the contact ;; location by the prior ar~ methods.
Note that the previously described calibration methods do not necessarily define a surface of interest which is coincident with the external surface of the display device.
When the user directs touches "through" the illuminated ,~ point of the screen from two or more different directions, the user may well be touching physically different points or.
the surface. Thus, when using a cathode-ray tube monitor ~ith a thick glass faceplate, the surrace of interest is ~ , '.~

.,, ~ ~ :
: - : .

2~3~

- located ~Jhere the phos~hors appear to be; and this may ~e defined implicitly in the application through the effect of - - ~ th-e- cal-ihration prQceduLe, allowinq the device to r~roJect a virtual response surface to match. The point of intersection of the thrust line with this response surface is closely approximated b~ finding the point within the virtual response surface having the least magnitude of the three-dimensional torque vector.
The method o the i..vention is distinguished from the prior art in that it reports the location within a surface of interast at ~,Jhich the three-dimensional torque vector takes on minimum magnitude (i.e. has shortest length).
Indeed, this may be taken as a description of the sole calculational requirement of the invention.
Further modifications will also occur to those skilled in ~his art including, for example, other types of sirnilarly functioning springs and sensors, as desired; and other locatlons of the external force-sensing platform contacting or abutting or other~ise touching the monitor or other apparatus carr~lring the surface upon i"hich touch or other contact events are to be locateà, including for more generai applications, placement intornall.y ot or o-hind tne .

- ~. .

--; ~ , ,i ,.
..

3 ~ ~

support2d apparatus. ~11 sucn supporting or abutting platforms or objects for a sur ace of interest of whatever shape, are generically termed herein as a "display surface portion" or similar term for generically termed "display apparatus". Other, more general "surfaces of interest" may be imagined. Consider the force applied to a glass window on a vending machine when a customer presses and points to a desired object far inside. The surface of each object is then a "surace of interest", potentially intersecting the line of minimum torque. A medical student, for e~ample, may point to invisible organs within the torso of an opaque human model, the rnodel may be quite hollow, ~et mathematical descriptions may be Eound Eor the surfaces of interest corresponding to the organs that would be there were the ~.
model real. Those s~illed in the art will readily see how such surfaces might be adequately descri~ed within the overall calculating means to be emploved, and how the method of the invention can be applied to such and other particular cases. Thus the "surface(s) of interest", and the corresponding desired "virtual response surface(s)", are defined by the application and the intent of the user, without being restricted by the specific illustrated .~

. . .

:
... , ~ ~

3 ~ ~

particulars of the desc-ibed mode of the invention. It i5 also to be understood, as before statedl that the invention is usefui with other types of electro-optical displ-ay sur,aces than catAode-ray tubes, including, but not limited to, LCD and LED displays. Such and other indicated modifications are deemed to fall within the spirit and scope of the invention as defined in the appended claims.

- , .
' , ', .~ ' ' ' ~ '' , ' ,~ ~ ', '.
., ' .
.'. ~

Claims (20)

1. A method of determination of touch location on a display surface portion of display apparatus, that comprises, contacting the display apparatus against a force-sensing platform disposed remotely from the display surface portion itself and having sufficient degrees of freedom and sensing sensitivity collectively to sense, in response to the thrust of touching a point of the display surface portion, the coordinate components of the resulting thrust vector and the components of the accompanying torque vector; calculating from the sensed forces a location on the display surface portion referred to which the measured force has substantially minimum torque magnitude; and outputting the resulting remotely sensed location as an estimate of said touching point.

2. A method as claimed in claim 1 and in which the force-sensing is responsive to six degrees of freedom of rigid motion of the display apparatus including the x, y and z axis coordinate components of the resulting thrust vector, and the roll, pitch and yaw components of the accompanying torque vector caused by said touching.

3. A method as claimed in claim 2 and in which said sensing is effected by spring mounting action between a pair of spaced parallel plates comprising the platform, and contacting the display apparatus at a region remote from the display screen portion thereof.

4. A method of determination of touch location on a surface carried by an apparatus, that comprises, contacting a region of the apparatus remote from said surface against a force-sensing means having sufficient degrees of freedom and sensing sensitivity collectively to sense, in response to the thrust of touching a point of the surface, the coordinate components of the resulting thrust vector and the components of the accompanying torque vector; calculating from the sensed forces, a location on a line of minimum torque to provide an intersection of the surface; and outputting the resulting remotely sensed location as an estimate of said touching point.

5. A method as claimed in claim 4 and in which said calculating comprises developing from the sensed forces an electrical model containing a direction perpendicular to the plane containing the thrust and torque vectors at a reference point, proceeding along said direction a distance equal to the magnitude of the torque vector divided by the magnitude of the thrust vector to reach a location on a thrust line closest to the reference point; and extending the thrust vector through said point to provide an intersection of the display surface.

6. A method of determination of touch location on a display surface portion of display apparatus, that comprises, contacting the display apparatus against a force-sensing platform disposed remotely from the display surface portion itself and being responsive to all six degrees of rigid motion of force and torque vectors resulting from touching said display surface portion of the display apparatus; sensing at the platform the forces developed by such touching of the display surface portion; calculating from the sensed forces a location on the display surface portion referred to which the equivalent of the sensed forces has substantially minimum torque magnitude; and outputting the resulting remotely sensed location as an estimate of the touching point.

7. A method as claimed in claim 6 and in which said calculating includes computing the point of least magnitude of the three dimensions of the said torque vector from among all points within the display surface.

8. A method of determination of touch location on a surface of an apparatus, that comprises, contacting the apparatus against force-sensing means disposed remotely from the surface itself and responsive to all six degrees of rigid motion of force and torque vectors resulting from touching said surface;
sensing at the said force-sensing means the forces developed by such touching of the surface;
calculating from the sensed forces a location on said surface referred to which the equivalent of the sensed forces has substantially minimum torque magnitude; and outputting the resulting remotely sensed location as an estimate of the touching point.

9. A method as claimed in claim 8 and in which the said apparatus comprises a computer input device, and said touching is effected by one of the hand of a user, a stylus or other tool.

10. A method as claimed in claim 8 and in which the computer input device includes a computer display device.

11. A method as claimed in claim 8 and in which said surface is a virtual response surface effectively coincident with a visible displaying surface.

12. A method as claimed in claim 11 and in which the virtual response surface is effectively coincident an exterior physical surface overlying the said visible displaying surfce.

13. A touch screen sensing apparatus for a monitor provided with a display screen having, in combination, force-sensing platform means contacting the monitor at a region thereof remote from the display screen; the platform means being provided with sensor means sufficient in number and degrees of freedom collectively to sense, in response to the thrust of touching a point of the display screen, measurements corresponding to the coordinate components of the resulting thrust vector and the components of the accompanying torque vector; means for calculating from said sensed forces a location on the display screen referred to which the measured force has substantially minimum torque magnitude;
and means for outputting the resulting remotely sensed location as an estimate of said touching point.

14. Apparatus as claimed in claim 13 and in which the sensor means provides six degrees of freedom, sensing linearly independent combinations of the x, y and z coordinate components of the thrust vector and the roll, pitch and yaw components of the accompanying torque vector.

15. Apparatus as claimed in claim 14 and in which the monitor rests upon the platform means.

16. Apparatus as claimed in claim 15 and in which the platform means comprises a pair of spring mounted spaced parallel plates.

17. Apparatus as claimed in claim 13 and in which the calculating means comprises a microprocessor provided with means for developing from the encoding an electrical model containing a direction perpendicular to the plane containing the thrust and torque vectors at a reference point, means enabling proceeding along said direction a distance equal to the magnitude of the torque vector divided by the magnitude of the thrust vector to reach a location on a thrust line closest to the reference point and for extending the thrust vector through said point to provide an intersection of the display surface.

18. Apparatus as claimed in claim 15 and in which means is provided for insuring an accurate and reproduceable positioning of the monitor upon the platform means.

19. Apparatus as claimed in claim 17 and in which the force-sensing platform means is provided with a plurality of distributed sensors positioned and of directional sensitivity sufficient to respond to all six degrees of freedom of platform motion by corresponding sensor outputs.

20. Apparatus as claimed in claim 19 and in which means is provided for respectively detecting, scaling and multiplexing the said sensor outputs and applying the same through an A/D converter, as of the slope type, to the said microprocesser.