CN103439030B - Texture force measuring method in a kind of haptic feedback - Google Patents

Abstract

The invention discloses the texture force measuring method in a kind of haptic feedback, consider the translational speed in objective grain surface factor and staff heuristic process and active pressing force, texture power is decomposed into vertical direction acting force and horizontal direction acting force.Wherein, vertical direction acting force is divided into objective factor acting force and subjective factor acting force, and objective factor acting force is determined by the grain surface micro-profile height of surveying and surface rigidity characteristic.Change in view of the normal acceleration in contact process can reflect the situation of change of staff moving process medium velocity and pressure, subjective factor acting force is the function of normal acceleration, and normal acceleration during emulation is the linear interpolation function of actual measureed value of acceleration under different pressures and speed.Horizontal direction texture power can be the frictional resistance that probe streaks grain surface, is determined by vertical direction pressing force and texture materials kinetic friction coefficient.

Description

Texture force measuring method in a kind of haptic feedback

Technical field

The invention belongs to reproducing texture force touch field, relate to a kind of power haptic rendering method of texture.

Background technology

At present, texture force touch expression technology is mainly set up power touch model when contacting with grain surface according to the contour feature of grain surface and reproduces out by power sense of touch by haptic feedback device.The modeling method expressed based on the texture of haptic device device is mainly divided three classes: the power haptic rendering model setting up texture based on geometrical constraint and physical model; The contour feature extracting grain surface from image sets up power haptic rendering model; The profile of actual measurement real-texture or vibration information, reproduce texture information by virtual reality technology.

Surveying instrument based on the texture power expression technology of actual measurement model is divided into two kinds: a kind of is contactless, as three-dimensional optical surface profiler, obtains surface three dimension elevation information by optical characteristics.This method to the cleanliness of measured surface and flatness requirement higher.Another kind is contact, is generally that the mode rapped by probe or press scans grain surface, and mutual by sensor record in status information.The subject matter of this method is " indeterminacy ", adds sensor and can change original connection status and dynamic perfromance, cause the deviation between measurement result and original case in system.

For the measurement based on probe scanning grain surface contact, domestic technique generally obtains the force value of probe scanning grain surface by force snesor, force value is directly expressed as texture micro-profile height as device for force feedback power output through simple process, in foreign study, the acceleration signal of the grain surface of measuring probe scanning simultaneously on basis at home, exports the validity improving texture sense of touch through conversion.

At present, having of haptic feedback is applied to: the proposition SensAblePHANTOM such as Hari laterally drag on the surface at a real-texture based on probe measurement grain surface information, estimate the shift perturbation perpendicular to surface direction simultaneously, and establish a texture power model on this basis; JochenLang etc. propose and use WHaT wireless touch sensor to measure power when handheld probe raps touch texture and acceleration, and calculate roughness and the rigidity of texture by these measurement results; The proposition attenuated sinusoidal signal such as Okamura describe the acceleration profile of probe when grain surface slides; The acceleration profile value that Guruswamy etc. propose to use iir filter to produce is multiplied by zoom factor as texture power; XianmingYe etc. propose with two strainometers and a force sensor measuring three-dimensional force signal, utilize normal force and tangential force to obtain the friction force of texture, and when measuring contact probe with piezoelectric membrane axial contact stress dynamic reading and for the haptic feedback of texture; Joseph etc. propose to carry out LPC filtering to the acceleration signal collected, and are exported by computer sound card by the acceleration signal of prediction, drive the voice coil motor on skin texture detection device to realize virtual texture haptic feedback by current amplifier.

Summary of the invention

Technical matters: the present invention proposes and a kind ofly utilize acceleration signal to carry out texture force modeling and texture force measuring method in considering the haptic feedback that the role of subjective intentions of people affects texture haptic display.

Technical scheme: the texture force measuring method in haptic feedback of the present invention, comprises the following steps:

In virtual texture power model, the virtual probe in the lever Controlling model of haptic feedback equipment is utilized to streak on dummy object surface, whether detection virtual probe and dummy object come in contact simultaneously, and whether have relative motion, if virtual probe and dummy object come in contact or without relative motion, then virtual texture power output is 0, otherwise, then calculate virtual texture power according to the following formula and export:

${\stackrel{\→}{F}}_{\mathrm{con}}={\stackrel{\→}{F}}_{\mathrm{ver}}+{\stackrel{\→}{F}}_{\mathrm{hor}};$

Wherein, for the vertical direction acting force of virtual texture power, for the horizontal direction acting force of virtual texture power, ${\stackrel{\→}{F}}_{\mathrm{ver}}={\stackrel{\→}{F}}_{\mathrm{obj}}+{\stackrel{\→}{F}}_{\mathrm{sub}};{\stackrel{\→}{F}}_{\mathrm{hor}}=\stackrel{\→}{f},$ Wherein, for objective factor acting force, for subjective factor acting force, horizontal direction friction force suffered by virtual probe during relative motion;

Objective factor acting force calculate according to the following formula:

${\stackrel{\→}{F}}_{\mathrm{obj}}=k_1\×k_s\×H\×\stackrel{\→}{n};$

Wherein, k _sthe constant describing texture materials stiffness coefficient, when H is virtual probe perception texture and the grain surface profile elevations h corresponding to contact point of grain surface, k ₁constant factor, it is vertical direction vector of unit length;

Constant k _sbe numerically equal to the linear coefficient A of vertical direction acceleration mean square deviation and pressing force, the computing formula of vertical direction acceleration mean square deviation is:

wherein, N is the group number of measured vertical direction acceleration, a _ibeing by the vertical direction acceleration information of a skin texture detection actual measurement, in order to obtain the linear coefficient A of vertical direction acceleration mean square deviation and pressing force, using correlation analysis and regretional analysis to calculate with pressing force F _p, sweep velocity v is independent variable, with vertical direction acceleration mean square deviation a _rmsfunctional relation for dependent variable: a _rms(F _p, v)=A × F _p+ B × v, wherein, A is vertical direction acceleration mean square deviation a _rmswith pressing force F _plinear coefficient, B is vertical direction acceleration mean square deviation a _rmswith the linear coefficient of sweep velocity v, F _pfor pressing force in actual measurement process, v is sweep velocity in actual measurement process; Constant factor k ₁computing formula be wherein, F _maxfor the maximum power output of actual forces feedback device, constant c ₁value is the arbitrary numerical value in 0.6-0.8;

Subjective factor acting force calculate according to the following formula:

${\stackrel{\→}{F}}_{\mathrm{sub}}=k_a\×a_{\mathrm{ver}}(F_p,v)\×\stackrel{\→}{n};$

Wherein, a _ver(F _p, v) be any pressing force size F in the perception using the method for linear predictive coding and bilinear interpolation to obtain _pwith the vertical direction acceleration under any perception velocities v, k _afor degree of will speed up signal a _ver(F _p, v) be converted into force signal constant factor, computing formula is:

Wherein, objective factor acting force is determined with subjective factor acting force the constant c of weight ratio ₂value is the arbitrary numerical value in 0.1-0.4, F _objfor the vertical direction acceleration of measured objective factor acting force under any pressing force and exploration speed; it is vertical direction vector of unit length;

Described horizontal direction frictional resistance calculate according to the following formula:

$\stackrel{\→}{f}=\mathrm{\μ}\×\left|F_{\mathrm{ver}}\right|\×\stackrel{\→}{t}$

Wherein, μ is the coefficient of sliding friction of virtual probe when virtual texture surface sliding, F _verfor texture power vertical direction acting force, for horizontal direction vector of unit length.

Beneficial effect: the present invention compared with prior art, has the following advantages:

1, use the acceleration signal of actual measurement to carry out the calculating of objective factor acting force, objective factor Calculation of the force formula is: this formula embodies the vital role that the objective characteristic of texture calculates virtual texture power, and wherein, parameter H embodies micro-profile information, the parameter k of grain surface _sembody texture materials stiffness characteristics.

Domestic technique generally only uses the micro-profile height H scanning and obtain to obtain the objective factor acting force of virtual texture power, and formula is: the present invention considers people in the rigidity information touching grain surface material in texture process on secondary basis, and the force capable haptic apparatus being fed back spend more can rediscover texture, and original formula adds parameter k _sas grain surface material stiffness coefficient, consider that soft, the hard Factors on Human of texture materials touches the impact of texture perception, add the validity of texture force touch perception.

2, propose one and consider that subjective factor is to the virtual texture haptic modeling algorithm of texture sensation influence, subjective factor Calculation of the force formula is: this formula embodies the impact of subjective perception custom on texture perception of people, parameter a _ver(F _p, v) for the present invention is based on the vertical direction accekeration under any pressing force and sweep velocity that actual measurement calculates, embody different perception custom and perception velocities to the impact of texture perception, k _aembody the weight relationship of subjective factor acting force and objective factor acting force, make the perception of virtual texture power more meet the actual impression of people.

At present, domesticly to there is no specifically to the research of people's subjective factor in perception.Foreign study worker has the investigation of subjective factor and has certain analysis, but does not also use vertical direction acting force to study subjective factor to the specific algorithm of reproducing texture force touch.

Accompanying drawing explanation

Fig. 1 is the logical procedure diagram of the inventive method;

Fig. 2 is the experiment schematic diagram of the inventive method;

Fig. 3 is the bilinear interpolation binary look-up table schematic diagram adopted in the inventive method.

Embodiment

Below in conjunction with embodiment and accompanying drawing, technical solution of the present invention is described in detail.

With reference to Fig. 1, texture force measuring method in haptic feedback of the present invention: the hand controller of steering force tactile sense reproduction equipment controls virtual probe convergence virtual texture surface, collision detection is carried out in virtual texture power model, before virtual probe and grain surface collide, the power output of power haptic apparatus is 0, when virtual probe and virtual texture surface produce collision rift, calculate the vertical direction acting force of fictitious force respectively with horizontal direction acting force vertical last, by hand controller by virtual texture power export.

With reference to Fig. 2, experiment schematic diagram of the present invention: when virtual probe contact grain surface, produce contact force be the acting force of vertical direction by distribution of contact forces with the acting force of horizontal direction be convenient to analytical calculation.

For 80 order sand paper, set forth specific embodiment of the invention process:

(1) use three-dimensional optical surface profiler KS-1100 to measure grain surface profile, obtain grain surface micro-profile height H;

(2) streaked in 80 order coated abrasive surface by skin texture detection pen, choosing pressing force F is respectively 0.2N, 0.5N, 0.8N, 1.1N, 1.4N, sweep velocity v is 50mm/s, 100mm/s, 150mm/s, 200mm/s, utilizes the acceleration transducer in detecting pen to measure corresponding vertical direction acceleration a respectively _i.Then can obtain the mean square deviation often organizing pressing force F vertical direction acceleration corresponding to sweep velocity v wherein, N is the number often organizing vertical direction accekeration, uses correlativity and regression analysis, obtains vertical direction acceleration mean square deviation a _rms(F _p, v) with pressing force F _pand the funtcional relationship of sweep velocity v: a _rms(F _p, v)=A × F _p+ B × v, wherein, A, B are constant coefficient.It is specific as follows that correlation analysis and regression are analyzed:

Correlation analysis is that the one studying different relationship between variables level of intimate commonly uses statistical method, sample correlation coefficient is called according to the statistic that can describe variable linearly degree of correlation that sample number calculates, usually, represent with Pearson simple correlation coefficient r, the computing formula of r is: wherein x _i, y _ifor the Two Variables sequence of correlation analysis, be respectively x _i, y _imean value, n is array x _i, y _iquantity.Respectively to the vertical direction acceleration mean square deviation a measured _rmsabout pressing force value F _pand sweep velocity v carries out correlation analysis, calculate correlation coefficient r, when | during r| < 0.3, for faint relevant; When 0.3≤| during r| < 0.5, be lower correlation; When 0.5≤| during r| < 0.8, for moderate is correlated with; When 0.8≤| r| < 1 is height correlation.

Confirming vertical direction acceleration mean square deviation a _rmswith pressing force value F _pand under the condition of sweep velocity v linear correlation, adopt the mathematical relation of multiple linear regression analysis determination correlative, the mathematical formulae describing this respective amount relation is called multiple linear regression equations.In multiple linear regression analysis, the estimation of regression coefficient adopts least square method.If multiple linear regression equations is: a _rms=β ₀× F _p+ β ₁× v, if multiple linear regression equation is:

wherein, for β ₀, β ₁estimated value, then residual sum of squares (RSS) is expressed as according to the principle of minimizing in differentio-integral equation, known SSE certainly exists minimal value, and in order to make SSE obtain minimum value, SSE is to β ₀, β ₁partial derivative should equal zero, that is:

$\left\{\begin{array}{c}\frac{\&PartialD;\mathrm{SSE}}{\&PartialD;{\mathrm{\&beta;}}_0}=-2\mathrm{\&Sigma;}(a_{\mathrm{rms}}-{\hat{a}}_{\mathrm{rms}})F_P=0\\ \frac{\&PartialD;\mathrm{SSE}}{\&PartialD;{\mathrm{\&beta;}}_1}=-2\mathrm{\&Sigma;}(a_{\mathrm{rms}}-{\hat{a}}_{\mathrm{rms}})v=0\end{array}\right.$

β can be obtained by solution formula ₀, β ₁estimated value then be constant factor A, B;

(3) virtual texture power vertical direction objective factor acting force modeling, formula is as follows:

${\stackrel{\&RightArrow;}{F}}_{\mathrm{obj}}=k_1\&times;k_s\&times;H\&times;\stackrel{\&RightArrow;}{n};$

Wherein, k _sthe constant describing texture materials stiffness coefficient, when H is virtual probe perception texture and the grain surface profile elevations h corresponding to contact point of grain surface, k ₁constant factor, it is vertical direction vector of unit length;

Constant k _sbe numerically equal to the linear coefficient of vertical direction acceleration mean square deviation and pressing force, be a _rms(F _p, v)=A × F _pconstant coefficient A in+B × v.Constant factor k ₁computing formula be wherein, F _maxfor the maximum power output of actual forces feedback device, constant c ₁value is the arbitrary numerical value in 0.6-0.8;

(4) virtual texture power vertical direction subjective factor acting force modeling, formula is as follows:

${\stackrel{\&RightArrow;}{F}}_{\mathrm{sub}}=k_a\&times;a_{\mathrm{ver}}(F_p,v)\&times;\stackrel{\&RightArrow;}{n};$

Wherein, a _ver(F _p, v) be any pressing force size F in perception _pwith the vertical direction acceleration under any perception velocities v, k _afor degree of will speed up signal a _ver(F _p, v) be converted into force signal constant factor computing formula be: wherein, objective factor acting force is determined with subjective factor acting force the constant c of weight ratio ₂value is the arbitrary numerical value in 0.1-0.4, F _objfor the vertical direction acceleration of measured objective factor acting force under any pressing force and exploration speed; it is vertical direction vector of unit length;

Any pressing force size F in said sensed process _pwith the vertical direction acceleration a under any perception velocities v _ver(F _p, computation process v) has used the method for linear predictive coding, specific as follows:

For normal acceleration value a (k) (k=1,2..., N) that actual measurement obtains, N is the number of one group of acceleration information under specific pressing force and sweep velocity, if vertical direction acceleration signal predicted value is wherein p is the exponent number of linear prediction filter, and the present invention gets p=10, for coefficient of linear prediction wave filter, obtained by Levinson-Durbin Algorithm for Solving, then the prediction residual of linear prediction filter then predicated error sequence E _r={ e (1), e (2), ..., e (k), ..., e (N) }, synthesize stationary random signal in order to white noise signal can be utilized, can set up a white noise sequence equal with its power spectrum density according to predicated error sequence, namely the power spectrum density of white noise sequence is then obtain with to pressing force F _i(i ∈ 1 ..., 5}) and sweep velocity v _j(j ∈ 1 ..., and 4}) corresponding parameter set up with pressing force F _p, sweep velocity v is the binary look-up table of coordinate, as shown in Figure 3, experiment measuring pressing force value F _i(i ∈ 1 ..., 5}) and scanning constant velocity amplitude v _j(j ∈ 1 ..., and 4}) corresponding point are defined as Q _ij=(F _i, v _j), this respective value is horizontal ordinate F _iordinate v _junder condition after the linear prediction of vertical direction acceleration signal wherein, corresponding pressing force F respectively _i(i ∈ 1 ..., and 5}) sweep velocity v _j(j ∈ 1 ..., 4}) under the coefficient of linear prediction wave filter of vertical direction acceleration signal with Power Spectrum of White Noise density σ ², defined function theoretical according to bilinear interpolation, the pressing force when perception texture and sweep velocity are no more than binary look-up table bounds and F ₁≤ F≤F ₅and v ₁≤ v≤v ₄time, to be evaluated on average obtained by the numeric weights of contiguous four net points, computing formula is:

$\left\{\begin{array}{c}g\left(R_1\right)=\frac{F_{i+1}-F}{F_{i+1}-F_i}g\left(Q_{\mathrm{ij}}\right)+\frac{F-F_i}{F_{i+1}-F_i}g\left(Q_{(i+1)j}\right)\\ g\left(R_2\right)=\frac{F_{i+1}-F}{F_{i+1}-F_i}g\left(Q_{i(j+1)}\right)+\frac{F-F_i}{F_{i+1}-F_i}g\left(Q_{(i+1)(j+1)}\right)\\ g\left(P\right)=({\stackrel{\&RightArrow;}{h}}_{\mathrm{new}},{\mathrm{\&sigma;}}_{\mathrm{new}}^2)=\frac{v_{j+1}-v}{v_{j+1}-v_j}g\left(R_1\right)+\frac{v-v_j}{v_{j+1}-v_j}g\left(R_2\right)\end{array}\right.,i\&Element;\{1,...,4\},j\&Element;\{1,...,3\}$

When pressing force when perception texture and sweep velocity exceed binary look-up table bounds, adopt boundary speed or terminal pressure value to solve current perception state g (P) as the sweep velocity value of this point or force value, its computing formula is as follows:

$\left\{\begin{array}{cc}({\stackrel{\&RightArrow;}{h}}_{\mathrm{new}},{\mathrm{\&sigma;}}_{\mathrm{new}}^2)=g\left(P\right)=\frac{v_{j+1}-v}{v_{j+1}-v_j}g\left(Q_{1j}\right)+\frac{v-v_j}{v_{j+1}-v_j}g\left(Q_{1(j+1)}\right)& F<F_1\\ ({\stackrel{\&RightArrow;}{h}}_{\mathrm{new}},{\mathrm{\&alpha;}}_{\mathrm{new}}^2)=g\left(P\right)=\frac{v_{j+1}-v}{v_{j+1}-v_j}g\left(Q_{5j}\right)+\frac{v-v_j}{v_{j+1}-v_j}g\left(Q_{5(j+1)}\right)& F>F_5\\ ({\stackrel{\&RightArrow;}{h}}_{\mathrm{new}},{\mathrm{\&sigma;}}_{\mathrm{new}}^2)=g\left(P\right)=\frac{F_{i+1}-F}{F_{i+1}-F_i}g\left(Q_{i1}\right)+\frac{F-F_i}{F_{i+1}-F_i}g\left(Q_{(i+1)1}\right)& v<v_1\\ ({\stackrel{\&RightArrow;}{h}}_{\mathrm{new}},{\mathrm{\&alpha;}}_{\mathrm{new}}^2)=g\left(P\right)=\frac{F_{i+1}-F}{F_{i+1}-F_i}g\left(Q_{i5}\right)+\frac{F-F_i}{F_{i+1}-F_i}g\left(Q_{(i+1)5}\right)& v>v_4\end{array}\right.$

Then obtain the correlation parameter of required point utilize power spectrum density generate white noise sequence { w (k) }, finally, utilize coefficient of linear prediction wave filter realize signal syntheses with white noise sequence { w (k) }, composite formula is: $a_g\left(k\right)=w\left(k\right)+\underset{l=1}{\overset{p}{\mathrm{\&Sigma;}}}{t}_l{a}_g(k-1)=w\left(k\right)+{\stackrel{\&RightArrow;}{h}}^T{\stackrel{\&RightArrow;}{a}}_g(k-1),$ A _gk () is the accekeration under required any pressing force and sweep velocity.

Claims (1)

1. the texture force measuring method in haptic feedback, it is characterized in that, the method comprises the following steps:

In virtual texture power model, the virtual probe in the lever Controlling model of haptic feedback equipment is utilized to streak on dummy object surface, whether detection virtual probe and dummy object come in contact simultaneously, and whether have relative motion, if virtual probe and dummy object come in contact or without relative motion, then virtual texture power output is 0, otherwise, then calculate virtual texture power according to the following formula and export:

Wherein, for the vertical direction acting force of virtual texture power, for the horizontal direction acting force of virtual texture power, wherein, for objective factor acting force, for subjective factor acting force, horizontal direction friction force suffered by virtual probe during relative motion;

Described objective factor acting force calculate according to the following formula:

Wherein, k _sthe constant describing texture materials stiffness coefficient, when H is virtual probe perception texture and the grain surface profile elevations h corresponding to contact point of grain surface, k ₁constant factor, it is vertical direction vector of unit length;

Constant k _sbe numerically equal to the linear coefficient A of vertical direction acceleration mean square deviation and pressing force, the computing formula of vertical direction acceleration mean square deviation is:

wherein, N is the group number of measured vertical direction acceleration, a _ibeing by the vertical direction acceleration information of a skin texture detection actual measurement, in order to obtain the linear coefficient A of vertical direction acceleration mean square deviation and pressing force, using correlation analysis and regretional analysis to calculate with pressing force F _p, sweep velocity v is independent variable, with vertical direction acceleration mean square deviation a _rmsfunctional relation for dependent variable: a _rms(F _p, v)=A × F _p+ B × v, wherein, A is vertical direction acceleration mean square deviation a _rmswith pressing force F _plinear coefficient, B is vertical direction acceleration mean square deviation a _rmswith the linear coefficient of sweep velocity v, F _pfor pressing force in actual measurement process, v is sweep velocity in actual measurement process; Constant factor k ₁computing formula be wherein, F _maxfor the maximum power output of actual forces feedback device, constant c ₁value is the arbitrary numerical value in 0.6-0.8;

Described subjective factor acting force calculate according to the following formula:

Wherein, a _ver(F _p, v) be any pressing force size F in the perception using the method for linear predictive coding and bilinear interpolation to obtain _pwith the vertical direction acceleration under any perception velocities v, k _afor degree of will speed up signal a _ver(F _p, v) be converted into subjective factor acting force constant factor, computing formula is:

Wherein, objective factor acting force is determined with subjective factor acting force the constant c of weight ratio ₂value is the arbitrary numerical value in 0.1-0.4, F _objfor the vertical direction acceleration of measured objective factor acting force under any pressing force and exploration speed; it is vertical direction vector of unit length;

Described horizontal direction frictional resistance calculate according to the following formula:

Wherein, μ is the coefficient of sliding friction of virtual probe when virtual texture surface sliding, F _verfor texture power vertical direction acting force, for horizontal direction vector of unit length.