Electronic Pitching Trainer and Method for Determining the True Speed of a Sports Projectile
A sensing device obtains range-related data—such as Doppler data or pulse time-of-flight data—from a sports projectile during flight. The time course of the range-related data is employed, in light of predictable characteristics of the projectile trajectory, to determine and output an accurate determination of the projectile speed for one or more points of interest in its flight. Such determination of speed may, for instance, be the speed at the time of projectile release, even though range-related data is gathered later in the flight, when the projectile is traveling neither so fast, nor straight at the sensor. Such sensing device may employ inexpensive short-range acoustic Doppler, and be incorporated into a target device that automatically displays to a pitcher-in-training the speed and ball/strike condition of each pitch.
This application is a divisional of U.S. patent application Ser. No. 13/206,735, filed on Aug. 10, 2011, entitled, “Electronic Pitching Trainer and Method for Determining the True Speed of a Sports Projectile,” which is a divisional of U.S. patent application Ser. No. 12/186,130, filed on Aug. 5, 2008, entitled, “Electronic Pitching Trainer and Method for Determining the True Speed of a Sports Projectile,” which claims priority from U.S. Provisional Patent Application Ser. No. 60/963,793, filed on Aug. 7, 2007, entitled “Electronic Pitching Training and Method for Determining the True Speed of a Sports Projectile,” all of which are hereby incorporated by reference herein.
BACKGROUNDThe trajectory of a sports projectile in free flight is characterized by steadily changing coordinates of position and velocity. Certain more limited characteristics of the full trajectory may be of special interest, however—notably, the component of the initial release speed in the intended direction of flight. This is what is typically meant when referring to the “speed” of a pitch in baseball, or of a serve in tennis, or of a slapshot in hockey. There is a need, therefore, for accurate and practical devices to measure such speeds.
In recent years, such measurements have been made somewhat more practicable by the development of the radar gun, wherein the rate of change of the projectile's distance from the device (its “range rate”) is taken from the Doppler shift of a reflected microwave signal. The release speed of the projectile may then be equated with the greatest range-rate measurement seen, as the speed of the sports projectiles of interest generally decrease monotonically after release.
Obtaining an accurate speed in this manner, however, is subject to difficulties and limitations. For instance, range-rate accurately reflects projectile speed only when the projectile is traveling straight at the measuring device. It may not be practical to meet this constraint if the intended path is not known in advance, or if the suitable locations for measurement are inaccessible, required for other uses, or out of the device's effective range. When the angle “theta” between the measuring line to the projectile and the projectile's line of travel is not zero, the measured speed will be diminished by a factor equal to the cosine of theta. The resulting error is sometimes known as the “cosine effect”.
In addition, a range-rate measurement will be reflective of release speed only if it is taken at the moment just after release. A baseball pitch, for instance, may lose ten miles per hour on its way to the plate, due to air drag—the exact amount depending on initial speed and pitching distance, among other things. Late detection of the pitch also complicates the effect of cosine error. Since the measuring device is unlikely to be perfectly located, the relevant theta is likely to increase with time, and the degree of cosine error will be greater, the later the device first “picks up” the pitch. Theta will also change with time due to a path curvature that is usually dominated by gravitational arc.
Thus a conventional radar gun may be difficult to use properly. Furthermore, a conventional radar gun of sufficient quality and sensitivity to give reliable and accurate readings when used properly can be an expensive device.
Although briefly discussed above in terms of the baseball pitch, analogous problems apply to measuring the speeds of other sports projectiles.
Therefore, there is a continuing need for an improved method for measuring the true speed of a sports projectile.
More particularly, there is a need for such a method in application to an inexpensive and convenient pitching trainer device. There exist passive targets for pitching practice, but these do not provide all the performance feedback desired. To monitor pitch velocity, an additional person must typically be dedicated to operating a separate speed-measuring device, such as a radar gun, and calling out the results. Pitch capture nets may provide some indication of ball/strike performance, but require hand tallying of all balls retrieved from both inside and outside the nets. Also, the mechanical design of existing targets can be deficient, with pitches of even moderate speed able to cause significant damage.
Therefore, there is need for a pitching trainer device that presents the speed of each pitch to the user without the user needing to leave his pitching position, and without tying up another individual in holding a radar gun or reporting speeds. There is also a need for such a device, without significant additional complexity, to be able to report and tally ball/strike performance. There is also need for such a device with improved robustness of design.
OBJECTSIt is an object of the invention to provide a method for determining the speed of a sports projectile that is not traveling substantially straight at the measuring device.
It is an object of the invention to provide a method for determining the release speed of a sports projectile based on data gathered later in its trajectory, when it has slowed.
It is an object of the invention to provide a method for determining the release speed of a sports projectile, based on data gathered by an inexpensive measuring device having limited range, operating in or near a target.
It is an object of the invention to provide an automated impact locating means for detecting and reporting which of one or more distinct target zones a sports projectile may have struck.
It is an object of the invention to provide one or more distinct target zones upon which impact may be detected by remotely disposed sensing, which sensing may thus be consolidated and protected.
It is an object of the invention to provide one or more distinct target zones which may be easily adjustable.
It is an object of the invention to provide one or more distinct target zones which may be provided in multiple alternatives that are both simple and inexpensive.
It is an object of the invention to provide an accurate and inexpensive method for determining the release speed of a baseball in the context of a pitching trainer device.
It is an object of the invention to provide a pitching trainer method and device that further incorporates a simple and inexpensive way to determine ball/strike performance.
It is an object of the invention to provide a pitching trainer method and device that provides pitch speed and ball/strike indications to the user immediately and automatically, without the user leaving pitching position, and without requiring the participation of another individual.
SUMMARYIn a particular form, devices and methods implemented in accordance with the invention may comprise a strike-zone target, backstop curtain, support structure, and an electronic sensing/display module with embedded firmware procedures. These together provide a practice and training environment for pitching. The sensing/display module provides to the user, at pitching distance, a display of the speed and ball/strike status of the last pitch. The sensing of pitch speed may be accomplished by means of continuous-wave ultrasonic Doppler. Preferably an efficient resonant piezoelectric emitter and a wide-range electret microphone provide excellent signal-to-noise ratio at low cost. The target and curtain are provided in such a form as to produce differing impact sounds. A microphone—which may be disposed at a convenient and protective distance from the target—picks up impact sounds that may be electronically differentiated into balls and strikes. The microphone for ultrasonic Doppler detection and for ball/strike discrimination may be the same microphone. The design allows all sensors and electronics to be separated from the larger passive elements. The resulting sensing/display module is a compact, easily detached unit that simplifies manufacture, user assembly, and protective storage during inclement weather.
Thus methods implemented in accordance with the invention integrate into one inexpensive, practical and convenient package the known passive functions of:
visible target definition, and
backstop
with novel automations of each of the active functions of:
umpire;
radar gun and operator;
tally function; and
immediate user feedback function.
In one important aspect, methods implemented in accordance with the invention comprise taking a range-related measurement at two or more times in the projectile's flight. These data are retained, along with their associated times of measurements. In one embodiment, a target exists in a known relation to the range-related measuring device, and a time of impact of the projectile upon the target may also be noted. From these data, parameters of a more complete trajectory are computed. The desired speed measurement is then computed with the aid of these parameters.
Broadly, the range-rate measuring device may depend upon any of several principles, such as, but not limited to: Doppler shift; increase in intensity of received energy radiated or reflected from the projectile; increase in angular subtense of a projectile image; and decrease in time-of-flight for a reflected pulse. In particular, it may respond to the Doppler shift of ultrasonic sound, radiated from the measuring device and reflected from the projectile. Whatever technique of range-rate measurement is employed, multiple measurement pairs comprising a range-rate value and a corresponding time-of-measurement value may be accumulated as the projectile nears the target. These data may thus sample a range-rate-versus-time curve of a particular shape. Such a history of range-related data may be used to implicitly identify plausible trajectories of a sports projectile, which trajectories are in turn associated with a narrow span of release speeds surrounding the true release speed. In this manner, an accurate release-speed estimate may be extrapolated to a value significantly greater than any of the range-rate measurements.
One embodiment employs a 40 kHz resonant emitter, along with a receiver comprising an electret microphone chosen for sensitivity in the 42-51 kHz range. Emitter and receiver may be located about 10 inches apart, and operated simultaneously. While waiting to detect and measure the range-rate of a pitched ball, the emitter may emit a constant, high-power sine wave at 40 kHz, which sound may, with the use of commonly available emitters, have an intensity of approximately 120 dB SPL at a distance of 1 foot. With use of appropriate combinations of filtering, amplification, and synchronous detection, the microphone output may be used to obtain a clean Doppler signal, neither overwhelmed nor degraded by the presence of a significant signal reflected from the back of a protective mesh window.
Although an ultrasonic emitter may be used for both emission and detection of ultrasound, as in pulsed time-of flight ranging, there may be difficulties in applying this technique to the ranging of a fast-moving object. The sound waves from a 40 kHz resonant emitter, for instance, return from a pitched baseball at a frequency roughly 3 kHz to 10 kHz higher, depending upon the speed of the pitch. As a sensitive resonant detector may offer a sensitive bandwidth of about 1 kHz, neither the emitter itself, nor even a differently tuned detector of otherwise similar design may be well suited for this task. Also, for the emitter to be used as detector, it must be pulsed off long enough to await and detect the reflected wave. For a fast moving object detectable only at close range, the required spacing of pulses may not provide appropriate and sufficient data points for subtle discrimination of range-rate changes.
Embodiments of the invention thus teach how a resonant emitter may be combined with a separate microphone of broader bandwidth capability, providing thereby an admirable capability to measure the Doppler range-rate of a sports projectile.
Electronics module 102 is seen supported from frame 104 by mounting strap 103. Mounting strap 103 may be of sheet steel, formed at an angle as shown in
Strike zone target sheet 107 provides a visible target to the user, and a distinctive impact sound to a transducer in electronics module 102. It may consist of 0.020″ thick polycarbonate, sized, for instance, for either a little league or a big league strike zone. It may be supported from frame 104 by elastic cords 109 running to grommets in its four corners. Multiple attachment holes in frame 104 for elastic cords 109 may provide the user with a choice of strike zone heights. Elastic cords 109 may pass around frame 104 from the front, before hooking into place in provided holes. Elastic cords 109 may then be of sufficient tension to hold target sheet 107 effectively flat. If backstop curtain 106 has been attached as described, it will tend to lie in the plane of the rear edge of the tubing comprising frame 104, while the target sheet 107 may lie in the plane of the front edge. Thus target sheet 107 may be held at a small spacing in front of backstop curtain 106, such that the sound of an impact may ring more freely in target sheet 107. At the same time, backstop curtain 106 may protect target sheet 107 by absorbing the bulk of a strike zone impact and preventing excessive deformation or corner stress in target sheet 107.
Frame 104 may be held erect by frame support 111 as seen in
Turning to
Turning to
Microphone 306 provides signal to preamplifier 402, generating as output the baseband signal 403. Also, through passive twin-T notch filter 404 the baseband signal 403 becomes the notched baseband signal 405. Bias and filtering circuitry 406 divides a regulated 4.0 volts 412 to provide 2.0 volts onto conductor 407, and also through buffer 408 onto conductor 409. This 2.0 volts is provided as the ground signal reference for the operational-amplifier circuitry. Baseband signal 403 typically carries a significant sinewave component at 40 kHz. To aid in impact detection and ball/strike discrimination, notch filter 404 is provided to suppress this, while leaving the audio frequency range largely unaffected. Notched baseband signal 405 may connect directly to a microprocessor A/D input for analysis of the substantial audio-frequency signal that may be generated when a baseball impacts strike zone target sheet 107, or backstop curtain 106.
Bias and filtering circuitry 406 also provides filtered and noise-reduced potentials near 0.0 volts and 4.0 volts for energizing microphone 306. Resistor 410 establishes the level of DC current flow through microphone 306, while the relatively large value of capacitor 411 diverts all AC current at frequencies of interest to the summing junction of preamplifier 402, and thence through current-to-voltage converting resistor 412. In contrast to common practice, resistor 412 is the first resistor to develop voltage from the AC current generated by microphone 306. Since resistor 410 is not used for this purpose, little or no AC voltage is developed across the microphone itself. Therefore, any feedback or shunt capacitances associated with the microphone output FET do not tend to reduce the microphone's high-frequency response.
-
- 1. A very low frequency or DC component that results from the synchronous rectification of the 40 kHz sinewave present on signal 502. This component may slowly vary in magnitude and sign, as the phase of the proximally reflected emitter output shifts.
- 2. A copy of the desired Doppler signal, now shifted down by 40 kHz. Thus the signal of interest here falls into the audio band, between, for instance, 2.5 kHz and 12 kHz.
- 3. An image signal, reflecting the presence of any acoustic or electronic noise originally present in the range of 28 kHz to 37.5 kHz. While no significant signal is normally to be expected in this range, any possible interference may be further reduced by judicious design of Doppler band amplifier 501.
- 4. A variety of much higher frequency components, including lower amplitude multiples of 40 kHz, higher amplitude multiples of 80 kHz, and multiple high frequency images of the Doppler signal.
Bandpass amplifier 506 is provided to selectively boost the desired Doppler signal, while strongly suppressing the low and high frequency components listed as 1 and 4, above. Signal Doppler 507 is then output at a level suitable to connect directly to a microprocessor A/D input, and need not be sampled at a rate higher than necessary for audio band signals.
Bandpass amplifier 506 may comprise three cascaded filter sections, each of the multiple-feedback bandpass type. The resulting transfer function has three zeros at the origin, and three pole-pairs. These pole pairs may offer a first frequency of 2.8 kHz, with Q of 6.3, and gain at 7 kHz of 7.0 dB; a second frequency of 5.4 k Hz, with Q of 2.35, and gain at 7 kHz of 4.0 dB; and a third frequency of 11.3 kHz, with Q of 7.5, and gain at 7 kHz of 12 dB.
Turning to
The LED display may be controlled by the 8 row outputs R0 through R7, and by the 10 column outputs C0 through C9. The 10 column outputs may encode, in parallel, the ON/OFF states of a logical row of 10 LEDs, while one of the 8 row outputs at any moment may be at logic high to indicate which row is to be so energized. Within MCU system 701, ancillary digital logic, such as decoders and shift registers, may be used to derive the 18 display drive outputs from a much smaller count of microcontroller chip pins. Diodes 708 operate with pull-down resistor 709 to generate an AND-OR logic of the operating switches and certain row outputs, such that if R0 is currently the high row, AND if switch 312 is closed, signal 705 will be read by the MCU system as at a logic high, OR if one of the other row outputs is high while its corresponding switch is closed, signal 705 will be read as high.
Turning to
-
- 1. In a first effect, the loss of ball speed due to air drag is clearly visible in 50 MPH curves 1201 and 1202. It is also important in 30 MPH curves 1204 and 1205, but is there overweighed and masked by a speed-of-descent effect, discussed next.
- 2. In a second effect, the lob-like pitches 1204 and 1205, thrown at only 30 MPH, are seen to show a range-rate that actually increases at points approaching the target. This is caused by the gathering downward speed of a ball on a descending arc. When the ball is first released, the range-rate measuring device may and should be at a height similar to the release height. Given the relatively large horizontal distance, a range-rate obtained at release time will, as desired, be little affected by any upward component to the release speed. By the time this reverses to a downward component, however, the ball may lie at a substantial angle above, and later below, the range-rate measuring device. This substantial angle converts the vertical component of velocity into something that affects the range-rate, further complicating accurate release speed measurement. In the case of high pitch 1204, this speed-of-descent effect causes late range-rate measurements to be elevated. In the case of low pitch 1205, the speed-of-descent effect causes the very late range-rate value to be depressed, becoming, in fact, significantly negative by the time of impact. Curves 1204 and 1205 have been plotted for low speed lobs so that the speed-of-descent effect may be visually obvious; the speed-of-descent effect, however, is present in the proximal range-rate data for pitches of all speeds.
- 3. In a third effect, all four pitch-curves may be seen to manifest a strong cosine effect drop-off before impact. For pitches that strike very close to range-rate sensors—in this case, high pitches—the effect may become significant only well after the range over which good data may first be collected. For most pitches, however, cosine effect must be accounted for in proximally measured data.
It is to be understood that the phenomena exhibited in
Turning to
The relatively simple method for selecting far points and near points just described has the advantage of looking for data at distances from the target that are reasonably independent of pitch velocity. Since there may be some unpredictable variation in data quality over short time intervals, it can be beneficial to have a second correction function available for use with a second set of nominal far point and near point distances, such as a nominal far point distance of 8 feet and a nominal near point distance of 4.7 feet. In other embodiments, other methods may be employed for selecting far points and near points. One approach may define two constant pre-impact times, and choose points for all trajectories at these times. For very fast pitches, however, such a method may select times when the ball is still so far from the target that good data are not yet available. Conversely, for very slow pitches, one fixed set of near and far point times may look for data from positions too close to the target. Multiple time pairs may be employed as a solution to this.
The Doppler value at a prespecified range-rate distance may be determined from measured data in a number of ways. The simplest may be to choose the Doppler value of that available data point having a range-rate distance closest to that desired. If data are free enough of noise and densely spaced in time, this method may suffice to assign Doppler values to a far-near pair. More precise results may be attainable by interpolating between the pair of points most closely bracketing the desired range-rate. Alternatively, more sophisticated methods may be employed to achieve more optimal results from limited and noisy data.
Conceptually, the goal is to fit the available data with an expected range-rate curve generated from the fewest possible free parameters. The desired characteristic of the trajectory, in this case the horizontal release speed, may then be computed from the parameters of the fit. Since only the final result is sought, there is great freedom in the way in which the parameters may be represented. For the purposes of the embodiment illustrated in
A particular application of a method implemented according to an embodiment of the invention will now be detailed. This both serves as a specific example of the more general principles, and illustrates how they may be applied in a system with limited computational resources.
We may represent a far-near pair as:
{{tFar, dopFar}, {tNear, dopNear}}
Where the first member represents the current best estimate of the far point. This is itself a pair, comprising “dopFar”, the current best estimate of the Doppler frequency of the far point, and “tFar”, the current best estimate of the time, relative to impact, that the ball passed this point. “dopNear” and “tNear” may be analogously defined for the second member, representing the current best estimate of the near point. Letting “t” vary over the time span found among the data points of the far set, we may, as follows, estimate the expected Doppler value “dopVal” as a function of “t” over this limited domain:
dopVal=dopFar*[1+Cf1*(t−tFar)+Cf2*(t−tFar)̂2] Equation 1
Where Cf1 and Cf2 are coefficients as computed below.
Similarly for the near set, we may approximate the Doppler curve as:
dopVal=dopNear*[1+Cn1*(t−tNear)+Cn2*(t−tNear)̂2] Equation 2
The following equations have been found to closely estimate the required coefficients:
Cf1=[Kf1+Kf2(1−dopNear/dopFar)]/tFar Equation 3
Cf2=[Kf3+Kf4(1−dopNear/dopFar)]/tFar̂2 Equation 4
Cn1=[Kn1+Kn2(1−dopNear/dopFar)]/tNear Equation 5
Cn2=[Kn3+Kn4(1−dopNear/dopFar)]/tNear̂2 Equation 6
Note that the four coefficients must be computed afresh for each cycle of estimation of each pitch, whereas the newly introduced values “Kf1” through “Kf4” and “Kn1” through “Kn4” represent fixed, pre-computed constants that are set by the details of the hardware and firmware design. At design time, these eight constants may be adjusted to provide best results on measured and computed data, such adjustment being accomplished by any of a number of known methods of parameter estimation, or by hand cut-and-try. This latter is not as difficult as it might at first seem, as the problem tends to break into four sub-problems with little interaction: refining Kf1 and Kf2 to best construct the slope of the far segment; refining Kf3 and Kf4 to best construct the curve of the far segment; refining Kn1 and Kn2 to best construct the slope of the near segment; and refining Kf3 and Kf4 to best construct the curve of the near segment.
Returning to Equation 1, it may be seen that when t is equal to the current estimate of the far-point time tFar, the computed estimate of the matching Doppler value must be exactly dopFar, the current estimate of its expected far-point value. The effect of the terms dependent on t, then, is to provide the slope and curve of a line passing through the far-point estimate. This line may then be compared to the actual data points of the far set, which fall at varying times before or after the far-point estimate. In particular, substituting for t the actual time of measurement of each point in the far set, predicted Doppler values may be derived from Equation 1. These may then be compared with the corresponding measured Doppler values, and the squared differences accumulated as a measure of the error of this approximation of the far segment. The current estimate of dopFar may then be adjusted as necessary to minimize this error. One simple and effective minimization of error may be performed as follows: Let a sum of all the measured Doppler values in the far set be called “sumFar”. After collecting a sum “sumFarFit” of all the Doppler values predicted for the far set by Equation 1, we may take:
dopFarImproved=dopFar*sumFarFit/sumFar Equation 7
Employing “dopFarImproved” in Equation 1 will now yield a sum-of-predictions exactly matching the sum of the measured Doppler values. In one very simple step this provides an error minimization that is very close to least-square. Similarly, we may take:
dopNearImproved=dopNear*sumNearFit/sumNear Equation 8
To refine tNear and tFar we may take:
tFar=10/V(dopFar) Equation 9
tNear=6/V(dopNear) Equation 10
where the function V returns the feet/second range-rate that is equivalent to the Doppler frequency provided as its argument and where the far-point and near-point range-rate distances are taken to be 10 ft. and 6 ft. respectively. These adjustments simply enforce the requirement that the far point and the near point fall at their corresponding range-rate distances.
A practical computation may then proceed as follows:
-
- 1. Scan the available data points, computing the range-related distance of each, and thereby establishing the far set and the near set.
- 2. Total up the Doppler values of the far set as sumFar, and of the near set as sumNear.
- 3. As initial estimates of dopFar and dopNear, take the Doppler values of the two measured data points with the range-related distances closest to 10 feet and 6 feet.
- 4. Set tFar and tNear in accordance with Equations 9 and 10.
- 5. Use Equations 3-6 to establish current coefficients for Equations 1 and 2.
- 6. Employing Equation 1 for all points in the far set, and Equation 2 for all points in the near set, accumulate values for sumFarFit and sumNearFit.
- 7. Replace dopFar and dopNear with the improved values computed in accordance with Equations 7 and 8.
- 8. Repeat from step 4, if further improvement is required.
- 9. In accordance with Equation 11, below, calculate pitch release speed from the final values of dopFar and dopNear.
We turn now to the second function, which may calculate pitch release speed “releaseSpeed” from the two trajectory parameters dopNear and dopFar. It has been found that a polynomial that is third-degree in dopFar and second-degree in dopNear provides a close estimate:
releaseSpeed=A1*dopFar+A2*dopFar̂2+A3*dopFar̂3+B1*dopNear+B2*dopNear̂2+C Equation 11
Again, the six constants—A1, A2, A3, B1, B2, and C—are precomputed, and are fixed by the hardware design, the intended conditions of use, and the way the two parameters provided describe the trajectory space.
In general, the derivation of an effective second function may begin by the collection of actual data with the device for which the computation is intended, under the intended conditions of use. These data may validate and be complemented by data generated by numerical simulation. Such simulation may require details of target and sensor location, projectile drag and drag-crisis characteristics, range of potential Magnus forces, intended flight distance, range of typical release points, operating frequency, and others. From a full range of required trajectories, typical data that are expected to be available in use may be prepared, and the true release speed of each data set matched with the trajectory parameters extracted by a previously established first function. The general form of a second function to compute the release speeds from the matching parameter values may then be proposed, and such constants as may be required to particularize this general form may be chosen to best fit the release speeds to the parameter determinations. If a very general functional form be used, such as a polynomial expansion, essentially all of the information in the fit may reside in the choice of these constant values. Such choices may be made automatically, using widely available software, such as the functions “FindFit”, “Fit”, and “NMinimize” found in Mathematica.
There is one further small simplification that may be made in the determination of dopFar and dopNear. In the name of expository clarity, this was not introduced earlier. The definition of tNear, and the parameter dopNear, may be modified as follows: with tFar defined and determined as before, tNear may simply be taken as a fraction, such a 6/10ths, of tFar. dopNear is then to be defined as the best estimate of the true Doppler value of the actual trajectory at this time. To effect this altered definition, Equation 10 is replaced by Equation 12:
tNear=0.6*tFar Equation 12
Also, the right sides of Equations 5 and 6 may reference tFar instead of tNear, and may employ different specific constant values:
Cn1=[Kn1′+Kn2′(1−dopNear/dopFar)]/tFar Equation 13
Cn2=[Kn3′+Kn4′(1−dopNear/dopFar)]/tFar̂2 Equation 14
This somewhat modifies the values that will be computed for dopNear, and modifies slightly the way dopNear and dopFar together describe trajectories. In turn, this results in somewhat different constant values being found for use in the second function. Various fixed fractions of the tFar estimate may also be used as limits to establish membership in the far and near sets, rather than examining each point separately for its range-rate distance. It is these slightly modified definitions that are assumed and presented in the firmware description to follow.
Firmware Structure of One EmbodimentDisplay state task 1505 is concerned primarily with the orderly progression of desired messages on the display, as initiated by events signaled from tasks 1504 and 1506, and as then dictated by the passing of time. If no new pitch is detected within, say 4 seconds of a preceding pitch, task 1505 may change the two character arrays from the display of the last pitch speed, to a sequential recitation of such accumulated statistics as the total pitch count, strike count, ball count, number of “walks” and number of “outs”. The number of walks may represent the number of times a four-ball count has accumulated before three strikes, and conversely for the number of outs, with the active strike and ball counts being maintained on LEDs 208, 209, and 210, and with these active counts being cleared when either strikes reach three or balls reach four. Each numerical statistic may be preceded by a two-character reminder and label for what follows, such as, for instance “PC” for pitch count, “S=” for the tally of strikes, “B=” for the tally of balls, etc. Each label or statistic may be presented for an appropriate short interval, such as second, before progression to the next. When the sequence is complete, task 1505 may return the last pitch speed to the display, but enter a battery-conserving mode in which this value is only flashed briefly every couple seconds. If there continues to be no pitch activity, the recitation of statistics may be repeated at intervals. At any time a pitch is detected by task 1504, the display state may be reset to display the new data, interrupting any statistics sequence.
There is a further, high-frequency display activity which must be steadily maintained. This comprises the advancement of the states of the display drive lines C0 through C9 and R0 through R9. If implemented at interrupt level, low-level multiplexing of the display may be achieved on steady and reliable 1-millisecond intervals. Such interrupt-level code may examine communicating variables stored by task 1505 to represent the currently desired display characters and balls/strikes/outs display. Since techniques of interrupt-driven activity are well understood in the art, opportunities for their effective use will be here noted in passing, but not detailed in flowcharts.
Switch monitoring task 1506 may maintain an image of the current switch and button states in continuously readable variables, and may signal switch and button transition events. Switch 312 may be employed to select an expected distance to the pitching rubber of 46 feet or 60 feet. This setting may then be used to choose the appropriate set of precomputed constants for pitch speed calculations. Switch 311 may be employed to select baseball or softball. Selection of softball may activate the application of an additional drag correction in calculating speed. Switch 310 may be used to select between two different sets of running statistics, such that a player #1 and a player #2 may alternate turns and be able to compare and compete. Changing the state of any of these switches may trigger special sequences to be displayed by task 1505, such that the user is prompted at such time as to which distance, ball type, and player are currently selected. Such a prompt sequence may also be triggered at power-up. Pushbutton 309 may be used to manually add to the ball count when a wild pitch has entirely missed the pitching trainer.
Turning to
If there are sufficient cycle widths for a Doppler determination, computation 1605 attempts to characterize these values with a single precise cycle width, as should be possible from input dominated by a sine wave of consistent frequency. While the Doppler signal sought will in fact be changing in frequency, this change will be too gradual to be significant over a small sequence of cycles. Computation 1605 may match the actual input transition times, as accumulated from successive widths, with a perfectly even sequence as predicted from a best-fit single width and starting phase. If the error in predicted transition times is seen small enough at 1606, the best-fit single width, and a system time characterizing it, may be a point entered into a list constituting the Doppler history. Whether or not a good Doppler value has been found from this cycle data, the older half of it may be discarded at 1608. This prepares for the next assessment of the cycle-width list to use half new data, and half old. Purging less than the entire list may improve the fraction of erratically noisy data that is able to contribute to good Doppler points.
At 1609, features of the developing Doppler history are examined. Points that are outliers in value or widely isolated in time may be discarded. If an increasing cycle width has been followed by a loss of ongoing data, this may be taken as a sign that a pitched ball has dropped below the Doppler field-of-view and that impact is immanent. If the signal amplitude of good data has grown very high, this may be taken as a sign that impact is immanent in the vicinity of the protective mesh window. Then if there is such a sign of possible immanent impact, and if there are at least, say, 9 good Doppler points in the history, state may be advanced to seekingImpact at 1610. Otherwise, collectingDopplerData will continue to look for sufficient good pitch data.
Within the scope of the method of the invention, many different choices may be made in the materials, form, or construction of distinct target areas, so as to render distinctive acoustic signatures to the impact of a sports projectile. In the particular application of the invention to the embodiment illustrated in
In one embodiment, the measures of lower and higher frequency activity may be accomplished as follows: The A/D sampling rate may be 48 kHz. As each sample becomes available, it may be summed separately into variables lowSum and highSum, and variables lowCount and highCount may each be incremented. When lowCount is found to have reached 25, the value of lowSum may be compared with a value previousLowSum, and the absolute value of the difference accumulated into a value lateLowerTotal. previousLowSum may then be set to lowSum, and lowSum set to zero. During the processing of each sample as it arrives, the value of highCount may also be examined, and if found to have reached 8, a similar set of operations may be performed to accumulate a value lateHigherTotal. The values of lateLowerTotal and lateHigherTotal accumulated by the end of the measurement period may then correspond, roughly but adequately, to the signal level in broad bands of frequencies peaking just over 700 Hz and just over 2 kHz, respectively. The computations just described entail acceptable compromises in accuracy of data use, in exchange for low computational demand. A further accommodation to possible speed limitation in processing may be made by performing at interrupt level the computations just described, representing the actions of code blocks 1803 and 1804. The portion of the checkingForStrike code invoked from main loop task 1504 may then monitor the results of these computations coming from interrupt level, and act on them as shown at 1806 to 1810. In this fashion it becomes less disruptive of measurement accuracy should code execution time between an exit at 1810 and reentrance at 1801 exceed one A/D sample time, as there may then be no overrun losses.
In certain embodiments, it may be desirable to use such dimensions and construction of frame 104 that the impact of a baseball on the frame may produce sounds similar in certain characteristics to those of an impact on target sheet 107. To prevent any such impact from registering as a strike, the parameters of the processing described above may be optimized to respond to such frame impacts as balls. An alternative embodiment, however, may include an additional sensor within electronics module 102. “Piezo benders” of known and inexpensive design are widely available. In addition to their typical use as acoustic emitters, they may also serve as sensitive vibration detectors. Such a device may be mounted to MCU board 304 so as to respond to vibrations of a few kHz in frequency passing to the module through mounting strap 103. Such vibrations may be present in substantial degree in the first few milliseconds following impact of a ball on frame 104, and absent after any impact confined to target sheet 107 or backstop curtain 106. Such vibrations may be detected, for example, if such an additional vibration sensor is connected between circuit ground and the center of divider 703. The firmware may then continue to use the DC value of the signal from divider 703 as a battery voltage indication, while checking for rapid fluctuations in this signal immediately after an impact is detected. If any such fluctuations are detected, the pitch may be reported as a ball. In this manner, the processing of the signal from microphone 306 may be optimized for distinguishing impacts on the two areas 107 and 106, based on their particular sound qualities only.
The actions taken in state reportingResult are detailed in
The routines findPitchSpeed( ) detailed in
The routine findPitchSpeed( ) is further detailed in
The routine improveDopNearAndDopFar ( ) is further detailed in
Most of the data handled in microcontroller system 701 may be stored and manipulated in 8-bit or 16-bit integer form. The computation of pitch speed, however, may be more readily and reliably coded for 32-bit floating point. The computation as described herein, however, is efficient of operation count. Software floating-point operations based on 8-bit integer arithmetic may be used, completing a pitch computation in about 50 milliseconds.
FURTHER DISCUSSIONThe method of the invention may be noted to apply to situations where range-related data may be available from only a single measuring position. It may thus be constraints inherent to the form of possible trajectories that are exploited to achieve data-sufficiency. This may be contrasted with more elaborate solutions, wherein multiple channels of data from sufficiently separated measuring positions may be employed for triangulation.
In other applications of the method of the invention, an impact time on a target at a known location may not be available. Thus, there may not be a common time to which the data and possible Doppler curves may be referenced. Thus in the absence of a known impact time, the central curve of
In particular, the method of the invention may be used with a three-parameter fit in application to a hand-held speed measuring device, or other speed measuring device not operated in conjunction with a target. A long-range device that it “picks up the pitch” immediately, for instance, may be used well to one side of the line-of-flight of a pitched ball. The method of the invention may then offer automatic cosine effect correction.
In further application of the method of the invention, extracted characteristics of a trajectory other than release speed may be used to advantage. Thus in the embodiment illustrated in
It is noted that certain physical phenomena that may in principle affect accuracy are not discussed in detail herein, as both simulation and experience show little effect on the pitch release speeds reported by the embodiments of the invention. These phenomena may include the Magnus force achievable with the degree of spin that even a talented pitcher may apply to a pitch; a degree of wind that is compatible with pitches landing where aimed; and to impact points covering the backstop curtain left-to-right, as well as high-to-low. The latter issue arises in principle, only because the cosine effect may interact somewhat differently with the speed-of-descent effect when the pitch misses to the side.
Claims
1. An automatic system comprising:
- a module comprising: speed identification means for identifying a speed of a sports projectile; speed indication means for providing an indication of the speed of the sports projectile using characters at least one inch in height;
- target means for receiving an impact of the sports projectile and for protecting the module from damage by projectile impact, the target means comprising: a transparent region through which the indication of the speed is displayed.
2. The automatic system of claim 1, wherein the module is attached to the target means.
3. The automatic system of claim 2, further comprising a flexible coupling means for isolating the module from projectile impact shock arising from the target means.
4. The automatic system of claim 2, wherein the module is detachable from the target means.
5. The automatic system of claim 1, wherein the module comprises all electronics necessary for its operation.
6. The automatic system of claim 1, wherein the transparent region comprises a protective mesh.
7. The automatic system of claim 1, wherein the target means further comprises at least one delineated target area, and wherein the module further comprises a target delineation means for indicating whether the sports projectile has impacted the at least one delineated target area.
8. The automatic system of claim 1, wherein the speed identification means and the speed indication means are contained within a housing which does not contain the target means.
9. The automatic system of claim 1, wherein the speed identification means comprises an acoustic sensor for generating a signal indicative of the speed of the sports projectile.
10. A method comprising:
- (A) at a module, identifying a speed of a sports projectile;
- (B) at the module, providing an indication of the speed of the sports projectile using characters at least one inch in height;
- (C) at a target means: (C)(1) receiving an impact of the sports projectile; and (C)(2) protecting the module from damage by projectile impact; (C)(3) displaying the indication of the speed through a transparent region of the target means.
11. The method of claim 10, further comprising, at the module, indicating whether the sports projectile has impacted at least one delineated target area of the target means.
12. The method of claim 10, wherein (A) comprises generating a signal indicative of the speed of the sports projectile.
Type: Application
Filed: Jul 8, 2013
Publication Date: Nov 7, 2013
Inventor: Jerry B. Roberts (Arlington, MA)
Application Number: 13/936,626
International Classification: G01P 15/00 (20060101);