Patent Application
20130068017
The present disclosure relates to apparatus and methods for analyzing the motion of a body and, more particularly, to apparatus and methods for orienting inertial sensors, which are used to determine the forces and moments acting on a body, along a single common plane.
This section provides background information related to the present disclosure which is not necessarily prior art. This section also provides a general summary of the disclosure, and is not a comprehensive disclosure of its full scope or all of its features.
Often in the analysis of movement of a body there is a need to determine the net forces and moments acting on the body. Whether it be during gait analysis of a human or animal, during the analysis of the throwing motion of baseball pitchers or the impact of sports equipment, or even during rehabilitation from injury or surgery, researchers can benefit from knowing the net loads that are placed on a subject's joints, body portions, or the equipment being used. Using a variety of methods, in some cases, these loads can be translated into individual muscle and passive tissue loading which has direct applications to the identification and prevention of injuries. Unfortunately, in the case of human or animal analysis, without invasive surgery, it is very difficult to directly measure the net joint forces and moments which are collectively referred to as the joint kinetics. However, given some information about the kinematics (acceleration, angular velocity and angular acceleration) of a subject, the joint kinetics can be deduced using inverse dynamics. These principles are equally applicable to analyzing the movement of equipment or bodies during impact, free flight, or other uses.
According to the principles of the present teachings, an apparatus is provided for analyzing movement of equipment that includes an inertial measurement unit continuously measuring six rigid body degrees of freedom of the equipment and outputting data representative thereof, wherein the inertial measurement unit includes a planar substrate define a single common plane. The inertial measurement unit further includes at least one angular rate gyro and at least one accelerometer sufficient to measure the six rigid body degrees of freedom and each being mounted on the single common plane. The apparatus further includes a communication device transmitting the data.
Further areas of applicability will become apparent from the description provided herein. The description and specific examples in this summary are intended for purposes of illustration only and are not intended to limit the scope of the present disclosure.
The drawings described herein are for illustrative purposes only of selected embodiments and not all possible implementations, and are not intended to limit the scope of the present disclosure.
Corresponding reference numerals indicate corresponding parts throughout the several views of the drawings.
Example embodiments will now be described more fully with reference to the accompanying drawings. Example embodiments are provided so that this disclosure will be thorough, and will fully convey the scope to those who are skilled in the art. Numerous specific details are set forth such as examples of specific components, devices, and methods, to provide a thorough understanding of embodiments of the present disclosure. It will be apparent to those skilled in the art that specific details need not be employed, that example embodiments may be embodied in many different forms and that neither should be construed to limit the scope of the disclosure. In some embodiments, aspects and/or features of recited embodiments can be combined into additional embodiments not specifically recited herein.
The terminology used herein is for the purpose of describing particular example embodiments only and is not intended to be limiting. As used herein, the singular forms “a”, “an” and “the” may be intended to include the plural forms as well, unless the context clearly indicates otherwise. The terms “comprises,” “comprising,” “including,” and “having,” are inclusive and therefore specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. The method steps, processes, and operations described herein are not to be construed as necessarily requiring their performance in the particular order discussed or illustrated, unless specifically identified as an order of performance. It is also to be understood that additional or alternative steps may be employed.
When an element or layer is referred to as being “on”, “engaged to”, “connected to” or “coupled to” another element or layer, it may be directly on, engaged, connected or coupled to the other element or layer, or intervening elements or layers may be present. In contrast, when an element is referred to as being “directly on,” “directly engaged to”, “directly connected to” or “directly coupled to” another element or layer, there may be no intervening elements or layers present. Other words used to describe the relationship between elements should be interpreted in a like fashion (e.g., “between” versus “directly between,” “adjacent” versus “directly adjacent,” etc.). As used herein, the term “and/or” includes any and all combinations of one or more of the associated listed items.
Spatially relative terms, such as “inner,” “outer,” “beneath”, “below”, “lower”, “above”, “upper” and the like, may be used herein for ease of description to describe one element or feature's relationship to another element(s) or feature(s) as illustrated in the figures. Spatially relative terms may be intended to encompass different orientations of the device in use or operation in addition to the orientation depicted in the figures. For example, if the device in the figures is turned over, elements described as “below” or “beneath” other elements or features would then be oriented “above” the other elements or features. Thus, the example term “below” can encompass both an orientation of above and below. The device may be otherwise oriented (rotated 90 degrees or at other orientations) and the spatially relative descriptors used herein interpreted accordingly.
As discussed above, inverse dynamics may form a part of the present teachings. Generally, inverse dynamics is based on link-segment modeling; a rigid-body method of approximating the dynamic behavior of the human body. The link-segment modeling technique breaks a human system into a collection of rigid segments linked by joints. All segments have known geometry, mass, center of mass, and inertia tensor as dictated by anthropometric data. Given a link-segment model of a portion of the human body, all that is needed to determine the net kinetics acting at the joints is information about the kinematics (linear acceleration along with angular velocity and angular acceleration) of the rigid segments. Then, according to the Newton-Euler equations (1 & 2 respectively), the net forces and moments acting on a segment can be determined.
Σ{right arrow over (F)}=m{right arrow over (a)} (Eq. 1)
Σ{right arrow over (M)}=d(I{right arrow over (ω)}/dt={right arrow over (ω)}×(I{right arrow over (ω)})+{dot over ({right arrow over (ω)}I (Eq. 2)
From these net forces and moments, which include the influence of the known weight of the segment, one can deduce the net joint force and moment by appropriately subtracting the influence of the known weight of the segment. There are a couple of different methods for determining the kinematical quantities appearing on the right-hand sides of Equations 1 and 2 which include the acceleration of the mass center of the segment ({right arrow over (a)}), the angular velocity of the segment ({right arrow over (ω)}) and the angular acceleration of the segment ({dot over ({right arrow over (ω)}), These methods, commonly referred to as motion capture, have varying degrees of accuracy, computational requirements, and applicability to studying natural human motion. They all provide the kinematic information needed to specify the right-hand sides of Equations 1 and 2.
The most popular method for recording human body kinematics is by using high-speed video, infra-red, or other cameras to track the three-dimensional position of a set of markers in space. The markers can be attached to the skin temporarily, rigidly attached to fixtures that are strapped tightly to various body segments, or (in rare cases) even attached directly to a subject's bones (e.g., by employing bone pins). The goal of tracking the markers is to develop an understanding of how the underlying bones are moving because their movement is what is approximated by the link-segment model. Unfortunately, there are several errors associated with each of these recording techniques that can skew the bone movement data. The errors are then magnified considerably by the four differentiation steps that are needed to solve for three kinematic quantities ({right arrow over (a)}′, {right arrow over (ω)}, {dot over ({right arrow over (ω)}) specified above. For example, forming the acceleration a requires two successive differentiations of the position data. Forming the angular velocity {right arrow over (ω)} first requires forming angular orientation measures (e.g. Euler angles, Euler parameters or similar) which themselves require differencing position data. The orientation measures must then be differentiated to obtain angular velocity. Next, the subsequent formation of angular acceleration {dot over ({right arrow over (ω)} requires a final differentiation of the angular velocity. Following these four derivative operations (and one difference operation), one can then solve for the net forces and moments acting on a segment per Equations (1) and (2). The differentiation operations are fraught with errors due to the presence of measurement noise and resolution limits of measuring the positions of the markers.
One of the most common problems that exists with skin-mounted markers is that there is usually movement of the soft tissue (i.e. skin, muscle, fat) between the marker and the bone. The small errors that result from this skin movement artifact are amplified considerably when the position data is numerically differentiated causing much larger errors in the kinetic predictions. The goal, for some researchers, has been to minimize this error. In some prior art systems, a number of different marker configurations are used during gait analysis to determine the optimal orientation for minimizing the effect of skin movement on position data. They tested constrained and unconstrained marker arrays, different attachment methods, and the effect of physical location of the arrays on the leg and found a particular combination of factors that was able to minimize the effect of skin movement. Some studies have used marker arrays that are physically connected to the bone however, these often affect movement patterns and therefore are good for testing the accuracy of less invasive (and more temporary) marker attachment methods, but are not often used to study motion on a larger scale. Methods have also been developed that use stereo radiography and more recently real-time MRIs to give very accurate descriptions of bone movement, however they expose subjects to radiation (stereo radiography) and prevent natural movement (MRI) due to many constraints associated with in-lab use. It should also be appreciated that many of these prior art solutions require cumbersome apparatus and systems for gathering of data.
One method that has been developed to try to circumvent the errors associated with skin movement is “markerless” motion capture. This method uses passive vision systems to capture human motion and then a computer model to determine the kinematics (and hence kinetics) of individual body segments. Some of the prior art systems employ a motion capture system that relies on the images from eight high-speed cameras to fit a “visual hull” to a person. The hull gives an approximation of the human form which, given some ideal assumptions, can allow for the position of the bones to be determined. This method theoretically reduces the effect of skin movement error on marker positioning. Other systems have been developed that employ a method for two-dimensional (sagittal plane) markerless motion capture in gait analysis. They fit a five-segment model to the contours of the lower leg for every frame of a recorded gait sequence using distance transformations. The model then supplied the position information of the lower leg segments without having to rely on skin-fixed markers. Unfortunately, this method was computationally inefficient at the time, requiring over four minutes to upload and fit the model to each frame of the video.
Another method adopted by researchers is to directly measure the kinematic quantities needed to solve for net joint kinetics using inertial sensors (accelerometers and angular rate gyros) It has been shown that accelerometers, along with position information for the accelerometers, can be used to determine all of the necessary kinematic quantities. More recently, a combination of accelerometers and angular rate gyros (a combination known as an inertial measurement unit or IMU) can be used to directly measure acceleration and rate of rotation in all three dimensions.
This technology means that net joint kinetics can now be determined with one numerical differentiation (of angular velocity to yield angular acceleration) instead of four differentiations of position and angle data in motion capture methods, thus substantially reducing error. It has been shown that when instrumented with an array of IMUs, the kinematic measurements closely matched those predicted by a well established video-based motion capture method. It is important to remember that with this technique, testing is no longer constrained to a laboratory setting, meaning that, for example, athletes can be instrumented during a game setting to collect data or during training or injury recovery. This ensures that movement is as natural as possible under testing conditions. Also, using IMUs to capture human movement substantially reduces the necessity for numerical differentiation which is the major source of error in employing the inverse dynamics methodology.
Most commonly an iterative solution to the Newton-Euler equations is used to determine the net joint kinetics. If only angular acceleration data is available, then the iteration starts at the most proximal end of the system being analyzed and progresses distally. Because the angular accelerations are usually determined with two numerical differentiations, this method produces noisy joint kinetic estimates. In an effort to reduce the noise, additional measurements of reaction kinetics are typically taken, and the analysis technique is altered to accommodate the additional equilibrium equations. Specifically, some methods employ a least-squares optimization technique to make use of the additional constraints placed on the system to improve the accuracy of the kinetic estimates.
Despite the efforts of researchers to minimize the errors associated with the kinematic inputs to an inverse dynamics analysis, it is still well recognized that these analyses are prone to errors. In fact, it has been determined that the five most common sources of error are as follows: (1) estimates of body segment parameters (i.e. geometry, mass, inertia); (2) segment angle calculations due to relative movement between surface markers and the underlying bone structure (skin movement artifact); (3) identification of joint center locations; (4) errors related to force plate measurements; and (5) motion marker noise and its effect on segmental accelerations. In some studies, it has been suggested that the approximation of joint centers as a possible source of error for inverse dynamics during gait analysis.
Several articles have focused on minimizing the number of these errors that they introduce into their analyses. For example, in some applications, in an attempt to reduce these errors associated with numerical differentiating position data, an iterative optimization approach is used where hip, knee, and ankle torques are varied until the joint angle time histories matched those measured using motion capture. Some have developed an EMG-based musculoskeletal model that could accurately predict net joint torques from EMG readings. This method also avoids many of the most common sources of error. Finally, it should be understood that the use of kinematics as captured with IMUs, avoid several of the common sources of error mentioned.
The present teachings provide apparatus and methods to determine the forces and moments acting on a single or multiple (connected) rigid bodies by exploiting measurements made possible by miniature inertial measurement units (IMU's).
The present teachings disclose how one may use these miniature IMU's for a new purpose; namely for the determination of the forces and moments applied to sports equipment, the forces and moments acting on major body segments, and, in general, the forces and moments acting on any rigid body or system of rigid bodies. Thus, these teachings concern how one can use miniature IMU's for determining kinetic quantities (forces and moments) in addition to or instead of the purely kinematic quantities considered in prior art concerning the use of miniature IMU's for sports and/or human motion analysis.
The teachings follows from data collected from miniature IMU's as input to inverse dynamic models which are then used to compute the resultant forces and moments acting on the rigid body (or bodies). The ability to determine the forces and moments acting on sports equipment and on major body segments has enormous implications for sports training, health and injury monitoring, prosthetic devices, human assistive technologies, robotics, and many other applications requiring the analysis of human movement. As one example, consider the sport of baseball and the potential injuries suffered by pitchers who develop excessive forces and moments across their elbow or shoulder joints. Consider also the benefit of knowing the force and moment created by a golfer's hands on a golf club during a golf swing, or the force/moment created by a bowler's hand on a bowling ball, etc. These teachings provide a highly portable, inexpensive and noninvasive way to quantify these reaction forces and moments at the shoulder, elbow, wrist, and fingers during live pitching, golfing, bowling, etc. The major competing technology for deducing these forces/moments is to employ motion capture cameras or (high speed) video capture. However, the present teachings provide significant advantages over camera-based systems, including greater accuracy, greater portability (ease of use), and lower cost.
Common to camera-based and IMU-based methods is the reliance on inverse dynamic models that relate basic kinematic quantities (acceleration, angular acceleration and angular velocity) to the forces and moments acting on a rigid body through the Newton/Euler equations of rigid body dynamics. As reviewed above, the methods differ substantially in how one arrives at the kinematic quantities for this purpose. Camera-based methods fundamentally begin by measuring the position of points on a rigid body. From these position measurements, one must then estimate angular orientation, differentiate position and angular orientation to yield velocity and angular velocity, and then differentiate those results again to yield acceleration and angular acceleration. The need for substantial (four) differentiations is a serious concern as each differentiation potentially magnifies the effect of measurement noise. By contrast, the IMU-based method of the present teachings fundamentally measures acceleration and angular velocity directly. One need only perform a single differentiation of angular velocity to yield angular acceleration to yield the data necessary for inverse dynamic models for the prediction of forces/moments. This method presents a major advantage that translates to greater accuracy for the computed forces/moments. A second major advantage is afforded by employing miniature, wireless IMU's. Doing so leads to a highly portable testing system that can be used in any environment (e.g., in a home or on the actual field of play) instead of in a specialized motion capture laboratory. A third major advantage is cost savings. The miniature wireless IMU's are orders of magnitude less expensive than traditional motion capture systems. The following examples provide further details on the collection and the analysis of kinematic data from equipment-worn and/or body-worn IMU's for analyzing kinetic quantities (forces and moments).
Shown in
{right arrow over (a)}
c
={right arrow over (a)}
p+{dot over ({right arrow over (ω)}×
in which {right arrow over (a)}p is the acceleration measured by the accelerometer, {right arrow over (ω)} and {dot over ({right arrow over (ω)} are the previously measured and computed angular velocity and angular acceleration, and {right arrow over (r)}c/p is the position vector locating the mass center of the lower leg relative to the position of the accelerometer. This position vector can be measured (or estimated) independently following the placement of IMU sensor board 120 on the lower leg using known or estimates of anthropomorphic data.
As an example of using the IMU-derived data for an inverse dynamics calculation, consider the simpler case of an IMU mounted in a bowling ball and then the computation of the net force and moment of the bowler's hand on the ball.
Application of Newton's second law requires
{right arrow over (F)}
hand
+{right arrow over (W)}
ball
=m{right arrow over (a)}
c (4)
where m denotes the known mass of the ball and {right arrow over (W)}ball is the known weight of the ball. Again, {right arrow over (a)}c denotes the acceleration of the mass center of the ball and it can be computed as shown in Eq. 3 above. Thus, one can readily solve for the reaction force of the bowler's hand on the ball from Eq. 4. In an analogous manner, Euler's second law about the mass center of the ball requires
{right arrow over (M)}
ball
+{right arrow over (r)}
g/c
×{right arrow over (F)}
hand
=I{dot over ({right arrow over (ω)}+{right arrow over (ω)}×(I{right arrow over (ω)}) (5)
where {right arrow over (r)}g/c denotes the position of the center of the grip (between thumb and finger holes) relative to the mass center of the ball and I is the centroidal inertia tensor of the ball (known from ball design data). Thus, one can now solve for the reaction moment of the bowler's hand on the ball from Eq. 5.
As one can imagine, it is of interest to people who study human movement to understand the forces that are placed on specific active (i.e. muscles) and passive (i.e. ligaments, tendons, etc.) tissues. Knowing these loads gives researchers the information necessary to diagnose the specific cause of injury and can also provide a means for preventing injury in both the short and long term. The issue with determining the forces applied by individual muscles is redundancy. The human body is full of redundant systems, meaning that there isn't only one muscle responsible for providing a certain force to a joint. The difficulty lies in determining the contribution of each of these redundant muscles to the net joint kinetics.
Researchers tend to employ static and dynamic optimization schemes to predict the contributions of individual muscles to the overall net joint kinetics. The theory behind these optimizations is to pick individual muscle forces that minimize some sort of function (i.e. sum of squared muscle force, or sum of total muscle work). Typically, the maximum muscle force is also somehow limited to provide another boundary condition for the solution. A static optimization usually relies on the net joint moments calculated through inverse dynamics and then solves the muscle redundancy problem at each step during time, so that once fully analyzed a complete time history of the individual muscle forces is created. In contrast, a dynamic optimization does not break the analysis into separate time steps. Instead, it relies heavily on joint-muscle models to determine individual muscle contributions during movement.
The difference between many of the studies that attempt to use optimization schemes is the quantity that they choose to minimize during their optimizations.
It should thus be appreciated that by employing the principles of the present teachings, namely measuring the kinematics directly using IMU's, errors associated with segment angle calculations due to relative movement between surface markers and the underlying bone structure (skin movement artifact); errors related to force plate measurements; and motion marker noise and its effect on segmental accelerations becomes much less significant compared to classical camera-based motion capture.
As first described in commonly-assigned U.S. Pat. Nos. 7,021,140 and 7,234,351, which are hereby incorporated by reference, it has been shown generally that inertial sensors 110 (e.g., MEMS accelerometers and/or rate gyros) can be used as the means to measure the rigid body dynamics of sports equipment, with some limitations.
By way of the present teachings, it will be more fully understood that with some changes these principles can be applicable to a wide variety of applications, including analysis of sports equipment, athletes, animals, patients, and the like. In other words, according to the principles of the present teachings, knowing the rigid body dynamics of the sports equipment or other equipment or member enables one to quantify equipment performance, athletic performance and/or general physiological properties. It should also be appreciated that the principles of the present teachings are equally applicable to equipment or members that are elongated, round, oval, or have any other shape. As will be explained, in some embodiments, the shape of the equipment or member can be, in some embodiments, irrelevant to acquisition of relevant data. It should be understood that the use of the terms “sports equipment” (or sports equipment 114) in the present application and in connection with preferred embodiments should not be regarded as limited to only sports equipment, but can include other equipment or bodies, such as rehabilitation equipment, training equipment, conditioning equipment, stress analysis equipment or other equipment useful in the determination of forces on a body, hand-held or hand-operated tools, devices, and equipment of all kinds, such as but not limited to construction; manufacturing; surgical; dental; and controls for aircraft, cars, weaponry, and the like.
In connection with the present teachings, the sensors, which output signals that are proportional to the acceleration (of a point) and the angular velocity of the moving body of sports equipment, can be used to compute useful kinematic measures of sports equipment and other equipment or body motion. By non-limiting example, such measures can include the velocity and orientation of the head of a golf club, the inclination of a baseball bat in the strike zone, the spin of a bowling ball, the dribble rate of a basketball, and the like. For many applications, the body of sports equipment may undergo an important transition from non-free flight dynamics to free flight dynamics as in, for example, the case of a baseball being thrown and released by a pitcher's hand. Similar transitions to free-flight arise in nearly all games employing balls such as football, tennis, golf, bowling, basketball, soccer, volleyball, to name a few examples, as well in other thrown launched objects such as javelins, discuses, shotputs, arrows, plastic discs (i.e. Frisbee®), etc. In these applications, the transition from non-free flight to free flight is of paramount interest for assessing athletic performance. In other words, skill in these sports in closely associated with achieving the correct “release conditions” as the object begins the free flight phase of motion.
In some embodiments, the present teachings can provide apparatus and methods for attaching inertial sensors 110 within or on sports equipment and simultaneously the means to protect such sensors from unwanted shock and vibration when in use.
A major challenge in accomplishing this goal is to attach or embed the sensors in sports equipment with a view towards protecting them from the impact forces that are regularly delivered to sports equipments during play. Impact may arise from contact with another piece of sports equipment (e.g. ball or puck with bat, racquet, club, stick, etc.) or with surfaces (e.g., ball, bat, racquet, club, stick on floor, wall, ceiling, fence, helmet, etc.) Such impact events produce extremely large forces (e.g., hundreds to thousands of g's when normalized by the weight of the equipment) that are of very short duration. The impacts generate structural waves in the equipment that typically propagate freely from the impact site to the site of sensor attachment. These impact-generated waves can severely degrade and/or permanently damage small inertial sensors 110 (e.g., MEMS scale, standard piezoelectric or any other type) being used for overall motion sensing (e.g., for measuring rigid body motion of the equipment). The sensors have limits to the shock they can reliably sustain before failing. Even below these limits, shock and vibration can lead to unwanted sensor output relative to the desired output associated with rigid body motion (i.e. the sensor output becomes the sum of that due to rigid body motion—desirable—and that due to impact-generated shock and vibration—undesirable).
Thus to succeed in achieving a system for sports training, it is desirable to properly isolate the sensors from unwanted, impact-generated shock and vibration. To this end, we have designed a family of sensor mounting concepts that attenuate shock and/or vibration inputs in sports equipment. These mounting concepts enable the inertial sensors 110 of the present teachings to survive harsh impacts (e.g., the impact of a golf club with a golf ball, a baseball bat with a baseball, a basketball with the floor, etc.). In some embodiments, the mounts employ compliant and dissipative materials (e.g. elastomers, foams, etc.) that significantly attenuate the shock and vibration that would otherwise be delivered to the sensors while in use.
A second challenge in deploying inertial sensors 110 on or in sports equipment is the need to align or to measure the alignment of the sensor axes relative to the major geometric features of the sports equipment. Consider the example of a golf club. A major indicator of skill in swinging a club is to achieve at impact the proper lie, loft and face angles of the club head. These three angles can be computed from sensor data (e.g., integrating the angular rate gyro data) using the algorithms set forth in the literature. However, to employ these algorithms one must first set or measure the three-dimensional orientation of the sensor axes relative to axes used to define the orientation of the club head. However, in many cases and in order to reduce sensor shock and vibration, it may be desirable to place the sensors not on the club head but on or in the shaft of the club and at the distal (grip) end. Then one must set or measure the orientation of the three axes that define the sensor system in the shaft relative to the three axes that define the orientation of the club head.
The principle of the present teachings covers two major solutions to resolve this challenge. One way is to achieve a prescribed orientation using mounting hardware that sets the orientation of the sensors. The second way is to measure the orientation actually achieved following mounting by in-situ calibration tests on the finished equipment. Once either method is employed for a golf club, one can then deduce the orientation of the club head from the data that describes the orientation of the sensors. This same need arises in all other applications whenever there is a need to know the position or orientation of a geometric feature. Examples include hockey (the orientation of the blade), baseball (the orientation of the bat label or the ball stitches), bowling (the orientation of the finger and thumb holes or center grip), and perhaps basketball (the orientation of the grooves or other ball features), and the like.
With reference to
The degree of isolation against shock is determined by the following two non-dimensional parameters:
The first parameter p denotes a non-dimensional ratio of the (short) time duration of the shock loading (ts) and natural period of oscillation of the mass and spring system (2π/√{square root over (k/m)}). For superior shock isolation, this ratio should be minimized which is achieved by having soft isolation material (i.e., isolation material stiffness k should be reasonably small). The second parameter ζ denotes the so-called damping ratio. For superior shock isolation, the damping ratio should be maximized (i.e., isolation material damping c should be reasonably large).
The degree of isolation against vibration is determined by the damping ratio ζ introduced in Eq. 7 above as well as by the non-dimensional frequency parameter
r=ω
excitation/√{square root over (k/m)} (Eq. 8)
in which ωexcitation denotes the frequency of a harmonic excitation. In this instance, vibration isolation is achieved whenever r>√{square root over (2)}. Moreover, superior vibration isolation follows by minimizing ƒ.
In practice, there are many materials and configurations of materials that can be effectively used in providing shock and/or vibration isolation for inertial sensors 110 mounted on or in sports equipment 114. Examples of these materials include various open-cell and closed-cell foams, rubbers, polymers (e.g. silicon rubber), cork, fluid- or gas-filled bladders, springs, and the like. The following examples serve as illustrations of effective designs for a variety of sports applications. These examples, which depict a wide range of mounting methods that incorporate shock and vibration isolation solutions, are illustrative and not comprehensive. One skilled in the art could incorporate similar elements in designs not included in the following.
In some embodiments, as illustrated in
With reference to
With reference to
Referring now to
With reference to
Still further, with reference to
As discussed herein, in some embodiment of the present teachings, an inertial measurement unit (IMU) sensor board 120 is used that contains inertial sensors 110 composed of accelerometers and angular rate gyros sufficient to measure the complete six degrees of freedom of the body of sports equipment 114. The inertial sensor 110 measures the acceleration vector of one point in/on the body as well as the angular velocity of the body. This can be accomplished by employing a single tri-axis accelerometer (alternatively, multiple single- or dual-axis accelerometers sufficient to measure the acceleration vector) and three single axis angular rate gyros (alternatively, fewer dual- or tri-axis angular rate gyros sufficient to measure the angular velocity vector).
Determination of the acceleration vector, representing three degrees of freedom, and the angular velocity vector, representing an additional three degrees of freedom, provides the means to compute the three-dimensional motion of the sports equipment 114 including the acceleration, velocity and position of any point associated with the body. In some embodiments, IMU sensor board 120 can transmit data to a central processing unit using wired or wireless transmission. In some embodiments, IMU sensor board 120 can comprise a microcontroller or microprocessor for data analysis (e.g. analog to digital conversion) and may also contain on-board memory storage (e.g. SD card or other memory storage device).
With particular reference to
As discussed herein, the ability to extend the present teachings to all shapes of sports equipment 114 rests on the ability to use IMU sensor board 120 to measure the translation and rotation of a frame of reference fixed to the body of sports equipment 114. Essentially, IMU sensor board 120 measures the translation and rotation of three-axes attached to or fixed in the body of sports equipment 114. This information can then be used directly to understand the translation and rotation of the body of sports equipment 114 and any feature on that equipment (e.g., position and orientation of laces, holes, markings, features on a sphere/ball).
In some embodiments, IMU sensor board 120 transmits data wirelessly to a receiver station that is connected to a laptop computer by a USB cable. Data collection software on the laptop can control the data acquisition and write the data to a file for subsequent analysis. The sensor data can now be used to deduce the acceleration, velocity and position (of any point on the ball) as well as the angular acceleration, angular velocity and angular orientation of the ball as described herein. Collectively, and in the context of bowling, these kinematic quantities provide significant information about ball motion that can be used for bowler training, ball fitting (drilling), and ball design.
With particular reference to
The principles of the present teachings are equally applicable in the determination and calculation of free flight dynamics. Moreover, the present teachings provide methods to distinguish and to analyze free flight dynamics from non-free flight dynamics of sports equipment 114 using the signals generated by inertial sensors 110 attached to or embedded in the sports equipment 114.
Briefly, as discussed herein, prior art covers the use of inertial sensors (e.g., MEMS accelerometers and/or rate gyros) as the means to measure the rigid body dynamics of sports equipment. The sensors, which output signals that are proportional to the acceleration (of a point) and the angular velocity of the moving body of sports equipment, can be used to compute useful kinematic measures of sports equipment motion. Example measures include the velocity and orientation of the head of a golf club, the inclination of a baseball bat in the strike zone, the spin of a bowling ball, the dribble rate of a basket ball, and the like. For many applications, the body of sports equipment may undergo an important transition from non-free flight dynamics to free flight dynamics as in, for example, the case of a baseball being thrown and released by a pitcher's hand. Similar transitions to free-flight arise in nearly all sports employing balls such as football, tennis, golf, bowling, basketball, soccer, volleyball, to name a few examples, as well in other thrown/launched objects such as javelins, discuses, shotputs, arrows, plastic discs (i.e. Frisbee®), etc. In these applications, the transition from non-free flight to free flight is of paramount interest for assessing athletic performance. In other words, skill in these sports in closely associated with achieving the correct “release conditions” as the object begins the free flight phase of motion.
The present teachings go beyond the prior art in that the teachings disclose 1) the means to readily distinguish the transition to free-flight using inertial sensors, 2) the means to analyze free flight dynamics and the release conditions at the start of free flight, 3) the means to do so with a simplified (less expensive) sensor configuration than that proposed in prior art, and 4) the means to correct for sensor drift errors by exploiting the free flight phase of motion.
The present teachings begin with the observation that the forces and moments acting on a body of sports equipment 114 abruptly change during the transition from non-free flight to free flight. For example, the forces and moments acting a bowling ball include 1) the (very large) force/moment system due to the bowler's hand, 2) the weight of the ball, and 3) the (exceedingly small) aerodynamic forces/moments. At the release of the ball, the forces/moments due to the bowler's hand are suddenly eliminated and this leads to large, detectable changes in the ball acceleration and angular velocity measured by embedded accelerometers and angular rate gyros. Following release, the ball is in free-flight and the acceleration of the mass center of the ball will remain constant (and equal to one g) and the angular momentum (hence angular velocity) of the ball will remain constant (assuming the excellent approximation that the weight of the bowling ball dominates the aerodynamics forces/moments). When the bowling ball strikes the lane, very large impact (normal) forces develop that again abruptly alter the measured acceleration and angular velocity via the embedded inertial sensors. Data presented in the following illustrates the abrupt transitions from non-free flight (ball in hand) to free-flight (release of ball and free flight) and back to non-free flight (ball impacts and rolls on lane) that are readily detectable by inertial sensors, how inertial sensors can measure the “release conditions” at the start of free flight and how this information can be used to support bowler training, ball/fitting drilling and ball design. Similar data is presented for an instrumented baseball. These two examples however are only representative of the many applications/sports to which this invention may be applied.
The prior art discloses the use of full six degree of freedom inertial measurement units (IMU's) to deduce the general (three dimensional) rigid body dynamics of moving sports equipment under general conditions without disclosing, teaching or otherwise distinguishing the special features associated with the transition from non-free flight to free-flight dynamics, how to evaluate free flight dynamics, and possible reduction of the sensors employed. For example, the prior art may teach the use of a three axis accelerometer and the simultaneous use of three single axis angular rate gyros all with sense axes aligned with a mutually orthogonal triad of directions attached to the body of equipment. Alternatively, the prior art may teach the use of multiple three axis accelerometers whose number are ultimately sufficient to deduce the full six degrees of freedom of the body of sports equipment (and with redundancies in this case). However, when the objective is to detect and measure the free flight dynamics of sports equipment 114, these prior designs are unduly complex and expensive relative to a simpler form of the invention disclosed below.
The simpler form of the present teachings follows from the fact that, under free flight, we already know the acceleration of one point on the rigid body for “free”; namely the acceleration of the mass center must remain 1 g. This tacitly assumes that the only external force acting during free flight is the weight of the body and this remains an excellent approximation to many bodies of sports equipment 114 whenever weight dominates aerodynamic forces. The new design employs just a single three axis accelerometer (alternatively three single axis accelerometers or one dual axis accelerometer used in conjunction with a single accelerometer) as the means to deduce the entire six degrees of freedom motion of the body without any need or use of angular rate gyros or any additional accelerometers.
With reference to
{right arrow over (a)}
P
={right arrow over (g)}+{right arrow over (ω)}×({right arrow over (ω)}×{right arrow over (r)}p/c) (Eq. 9)
where, {right arrow over (a)}p denotes the acceleration measured by an accelerometer at point p, {right arrow over (g)} represents the known acceleration of the mass center c, {right arrow over (ω)} represents the unknown angular velocity of the body, and {right arrow over (r)}p/c represents the known position of p relative to c. (Note that the angular acceleration is presently assumed to vanish during free flight on the assumption that no appreciable aerodynamics moment exists that would otherwise alter the angular momentum of the body. One could also relax this assumption and include the angular acceleration in the above relation. Doing so would lead to a set of three first-order differential equations to solve for the angular velocity in lieu of the simpler algebraic equations above). Thus, provided one knows the orientation of gravity {right arrow over (g)} relative to the sensor frame of reference, one can reduce Eq. 9 to three algebraic equations for solution of the three components of {right arrow over (ω)}. Knowledge of {right arrow over (ω)} then completes the knowledge of the entire six degrees of freedom of the body. This process becomes even simpler when the accelerometer also detects gravity (as is the case with MEMS accelerometers that measure acceleration down to dc) as described next. The important point is that by employing a single (three axis) accelerometer (and not angular rate gyros and/or any additional accelerometers), one can determine all the information needed to analyze the six degree of freedom rigid body motion of a body of sports equipment 114 in free flight.
A further simplification of the above theory results when employing a three-axis accelerometer that is capable of detecting gravity in addition to the acceleration (motion) of a point. Such accelerometers are common. For instance, common MEMS accelerometers (manufactured by Freescale Semiconductor, Analog Devices, Invensense, or Kionix, for example) measure down to dc and therefore detect gravity in addition to the acceleration of a point. For instance, if a MEMS accelerometer were fastened to a stationary table, the measured output of that accelerometer would be given by:
{right arrow over (a)}
measured
=g{circumflex over (K)} (Eq. 10)
Where g represents the local gravitational constant and {circumflex over (K)} is a unit vector directed vertically upwards from the surface of the table. Should the table now accelerate with acceleration {right arrow over (a)}, the measured output of the accelerometer would now change and become
{right arrow over (a)}
measured
g{circumflex over (K)}+{right arrow over (a)} (Eq. 11)
Now consider again the case where the accelerometer is mounted on or in a moving body of sports equipment 114 that undergoes a transition to free flight. As in
{right arrow over (a)}
measured
=g{circumflex over (K)}+{right arrow over (a)}
p
=g{circumflex over (K)}+[−g{circumflex over (K)}+{right arrow over (ω)}×({right arrow over (ω)}×{right arrow over (r)}p/c)] (Eq. 12)
Hence the measured output of the accelerometer reduces to
{right arrow over (a)}
measured={right arrow over (ω)}×({right arrow over (ω)}×{right arrow over (r)}p/c) (Eq. 13)
So in this common instance, the output of the accelerometer alone can be used to immediately deduce the three components of the angular velocity vector {right arrow over (ω)} and without additional knowledge of the orientation of gravity {right arrow over (g)} with respect to the sensor frame as previously required above.
To this end, let (î, ĵ, {circumflex over (k)}) be a triad of mutually orthogonal unit vectors fixed in the accelerometer and aligned with the three mutually orthogonal sense axes of the accelerometer as illustrated in
{right arrow over (ω)}=ωxî+ωyĵ+ωz{circumflex over (k)} (Eq. 14)
represent the unknown angular velocity vector having components (ωx, ωy, ωz) along (î, ĵ, {circumflex over (k)}), respectively. Similarly, let
{right arrow over (r)}
p/c
=xî+yĵ+z{circumflex over (k)} (Eq. 15)
be the position vector of point p relative to c having the components (x, y, z) along (î, ĵ, {circumflex over (k)}) respectively. Finally, let
{right arrow over (a)}
measured
=a
x
î+a
y
ĵ+a
z
{circumflex over (k)} (Eq. 16)
represent the measured output of the accelerometer having components (ax, ay, az) along (î, ĵ, {circumflex over (k)}) respectively. Substituting Eqs. 14, 15, and 16 into Eq. 13 and expanding into the three associated component (scalar) equations yields the following three algebraic equations for solution of the three unknown angular velocity components (ωx, ωy, ωz):
a
x
=−x(ωy2+ωz2)+yωxωz+zωxωz (Eq. 17a)
a
y
=−y(ωz2+ωx2)+zωyωz+xωyωz (Eq. 17b)
a
x
=−z(ωx2+ωy2)+xωzωx+yωzωx (Eq. 17c)
In summary, due to the special conditions arising during free flight, the measured outputs from a single, three-axis accelerometer (ax, ay, az) enables one to deduce the three-dimensional angular velocity of a body of sports equipment 114 by simultaneous solution of the three Eqs. 17a, 17b, and 17c. Doing so provides an indirect yet inexpensive means to evaluate the rotational dynamics of a body of sports equipment 114 (e.g. a ball) when in a state of free flight.
Conversely, if one employed solely a three axis rate gyro (or a combination of single or dual axis rate gyros) yielding the three components (ωx, ωy, ωz) of the angular velocity vector, then one can invert the process above and compute the acceleration of any point on the body of sports equipment 114 by evaluating the right-hand sides of Eqs. 17a, 17b, and 17c.
Thus, a single inertial sensor (a three axis accelerometer/equivalent or a three axis rate gyro/equivalent) capable of measuring just three degrees of freedom ((ax, ay, az) or (ωx, ωy, ωz), respectively) can also be employed in IMU sensor board 120 to deduce the remaining (unmeasured) three degrees of freedom of the body of sports equipment 114 via Eqs. 17a, 17b, and 17c when the body is in free flight.
In addition to the uses disclosed above, one can also use the special features of free flight dynamics of sports equipment 114 to correct for drift errors that frequently limit the accuracy of full (six degree of freedom) inertial measurement units (IMU's) that are introduced upon integrating the sensor data. Both accelerometers and angular rate gyros introduce drift errors which, when integrated over a period of time, produce inaccurate predictions of kinematical quantities including velocity, position and orientation (angles). The analysis of the free-flight dynamics of sports equipment 114 however provides constraints on the measured acceleration components (ax, ay, az) and the measured angular velocity components (ωx, ωy, ωz) in IMU's that enable drift correction as detailed next.
First, integration of the angular velocity components obtained from the angular rate gyros enables one to deduce the orientation of the body of sports equipment 114 with respect to the initial orientation (initial condition for the integration). Refer to
in which the rotation matrix R is found upon integrating the angular velocity vector measured by the body-fixed angular rate gyros. As a specific example, Euler parameters are used to determine the rotation matrix R of a golf club in King et al., “Wireless MEMS Inertial Sensor System for Golf Swing Dynamics,” Sensors and Actuators A: Physical, vol. 141, pp. 619-630, 2008. Computation of R then enables one to compute the acceleration of the mass center of the body of the body of sports equipment 114 with respect to the inertial frame (Î, Ĵ, {circumflex over (K)}) and then proceed to compute the velocity of the mass center (one integration) and the position of the mass center (a second integration) again with respect to the inertial frame. These integration steps (including the original integration leading to R lead to inaccuracies due to sensor drift. To correct for drift, one can exploit the known kinematics of the mass center for a body of sports equipment 114 in free flight. Specifically, the acceleration of the mass center deduced from the measured acceleration and angular velocity is constructed from
{right arrow over (a)}
c-constructed
={right arrow over (a)}
p+{right arrow over (ω)}({right arrow over (ω)}×{right arrow over (r)}c/p)={right arrow over (a)}measured+{right arrow over (ω)}×({right arrow over (ω)}×{right arrow over (r)}c/p)−g{circumflex over (K)} (Eq. 19)
with the resultant vector written with respect to the inertial frame by employing the rotation matrix R. Due to drift errors, this result will be close to but not equal to the true acceleration of the mass center given by −g{circumflex over (K)} (recall that {circumflex over (K)} is a unit vector directed positive upwards.) Let the error in the constructed acceleration be given by the vector
{right arrow over (ε)}={right arrow over (a)}c-constructed+g{circumflex over (K)} (Eq. 20)
Then substitution of Eq. 19 into Eq. 20 yields the following expression for the estimated error vector
ε={right arrow over (a)}measured+{right arrow over (ω)}×({right arrow over (ω)}×{right arrow over (r)}c/p) (Eq. 21)
wholly in terms of the measured outputs from the three axis accelerometer and angular rate gyro. One can now use Eq. 21 to estimate and correct for the error at any (or multiple) times during the free flight phase of motion of a body of sports equipment 114. Other drift correction algorithms are also possible that exploit the fact that, during free flight, the measured angular velocity and the measured acceleration are not independent as previously shown in Equations 17(a)-(c).
The above teachings have been reduced to practice in two example applications, namely bowling and baseball, as described next. Again, it is important to emphasize that these are just two illustrative examples, and that the teachings may be readily applied to all other sports or other equipment where the body enters a state of free flight.
As discussed herein, of paramount interest in the sport of bowling are the so-called “rev rate” and “axis point” of the ball as it is released from the bowler's hand. The rev rate is the magnitude of the angular velocity of the ball upon release and it is measured in the units of revolutions per minute (RPM). For instance, professional bowlers frequently achieve rev rates above 350 RPM. The axis point locates the “spin axis” or equivalently the direction of the angular velocity vector on the surface of the ball upon release. More precisely, the axis point is the point on the ball surface where the angular velocity vector pierces the ball (for a right-handed bowler). Taken together, the rev rate and axis point define the magnitude and direction of the angular velocity of the ball as it is released from the bowler's hand. These quantities, together with the release speed of the ball, will determine the degree to which the ball will “hook” in the lane thereafter. (In addition to these release conditions, the hook is also influenced by the oiled lane conditions down the lane). The ability to hook the ball consistently and to control the degree of the hook are the hallmarks of an expert bowler.
The transition to free flight and then lane impact is readily detectable in the sensor data as illustrated in
An important part of the present teachings concerns how one can determine the rev rate and axis point from the simplified sensor design (i.e. IMU sensor board 120) described above that employs just one three axis accelerometer (and no angular rate gyros). This simpler sensor may be far less expensive given the relatively higher cost of angular rate gyros. Nevertheless, following the present teachings described herein, this simplified design can provide excellent measures of the rev rate and axis point as demonstrated next in
As a second example, consider the photograph of
When using the full (six measurement degree of freedom) IMU, the data can be used to reconstruct the acceleration, angular acceleration, velocity, angular velocity, position and angular orientation of the ball as functions of time during throwing and free flight. These kinematic measures, and metrics derivable from these, have tremendous potential for use in player/pitcher training for baseball, softball, and any other sport involving the use of sports equipment 114 launched into free flight.
For instance, the results of
The transition to free flight of a baseball is readily observable in the angular velocity components and/or the acceleration components. As an illustrative example,
The foregoing description of the embodiments has been provided for purposes of illustration and description. It is not intended to be exhaustive or to limit the invention. Individual elements or features of a particular embodiment are generally not limited to that particular embodiment, but, where applicable, are interchangeable and can be used in a selected embodiment, even if not specifically shown or described. The same may also be varied in many ways. Such variations are not to be regarded as a departure from the invention, and all such modifications are intended to be included within the scope of the invention.
