Sie sind auf Seite 1von 11

A dynamic approximation of balanced gymnastics landings

Alison Sheets and Mont Hubbard


Department of Mechanical & Aeronautical Engineering, University of California, Davis, USA

Abstract
A gymnast can balance after an uneven parallel bar dismount by contacting the ground with an appropriate body position and angular velocity. Balanceable gymnastics landings were predicted by dividing landing into two parts: a short duration impact phase when footmat contact forces are high, and a subsequent longer balancing phase during which the gymnast exerts active control at the joints to attenuate the resulting post-impact velocities and achieve a desired motionless terminal configuration. This paper identifies controllable sets of pre-impact landing positions and velocities for a typical female collegiate gymnast that are feasible targets for the end of flight following a single- and double-backward somersault uneven bar dismount with forward mass centre (CM) velocity. Simple algebraic impulsemomentum models for the impact phase determine impulses necessary to stop the feet and slow angular velocity. After impact, a dynamic analysis is needed to determine stability because non-negligible post-impact angular and linear velocities are possible. For almost all successfully balanced landings, ground contact occurs with the CM positioned above the feet and between the toe and heel, because the linear impact impulses that stop the feet also contribute to an angular impulse that slows angular momentum. Only at the largest pre-impact angular velocities following a double-somersault dismount can contact occur with the CM in front of the toe. The gymnast is in balanceable preimpact orientations longer when rotating more slowly and with a smaller angular momentum; therefore balance may be more likely after performing a single-somersault dismount with a large moment of inertia.
Keywords: dynamic balance, gymnastics landing, impact, model

Introduction
The goal of an elite female gymnasts uneven parallel bars dismount is to release the high bar, complete

Correspondence address:
Alison Sheets Department of Mechanical & Aeronautical Engineering, University of California, Davis One Shields Avenue Davis, CA 95616-5294 USA Tel: 001 530 752 2163 Fax: 001 530 752 4158 E-mail: alsheets@ucdavis.edu

rotations and/or twists, and end balanced without moving the feet. To achieve balanced landings ground reaction forces (GRFs) of over 10 times body weight (BW) must be attenuated (McNitt-Gray, 1993) and angular velocity slowed from as high as 18 rad s1 to zero with the centre of mass (CM) between the toe and heel. This occurs over roughly one second, during which there is a short duration (t = 0.04 s) impact characterised by large peak footmat contact forces, and a longer subsequent balance phase with GRFs of approximately one BW (McNitt-Gray, 1991, Figure 1). During balance, the gymnast actively exerts joint torques while interacting with a compliant surface to reduce post-impact velocities and achieve a motionless

2007 isea

Sports Engineering (2007) 10, 209220

209

Dynamic approximation of balanced landings

A.L. Sheets

terminal configuration. Previous research has modeled impact and balance, and experiments have investigated landing as a whole. The models of drop landing impacts onto rigid surfaces must include wobbling masses if the details of GRFs, intersegmental joint torques, or passive tissue contributions to mechanical energy dissipation were of primary interest (Gittoes et al., 2006; Gruber et al., 1997; Pain & Challis, 2002, 2006). Following initial ground contact with the heels, a rigid model overestimated peak GRFs by 150% (Pain & Challis, 2006). Following a drop landing with forefoot then heel ground impact onto a rigid surface, peak forces were overestimated by as little as 13.9% using a rigid model (Gittoes et al., 2006). Additionally, the time to peak GRF using a rigid model more closely matched experimental data than did the wobbling mass model (Gittoes et al., 2006). Impact with a compliant surface rather than a rigid one would similarly decrease peak impact forces and increase the time to decelerate the mass (Nigg & Liu, 1999). Also, muscle pre-activation increases with landing surface compliance (Nigg & Liu, 1999). Because the gymnastics surface is compliant, muscles would be pre-activated and the body would be more rigid at impact. This further diminishes the effect of wobbling masses on GRF because the percentage of mass that wobbles decreases with increased muscle activation. While wobbling masses are needed in the previous models, gymnastics landings are forefoot then heel landings onto compliant surfaces. Furthermore, as is shown below, the details of the amplitude and duration of the GRF are not essential to the study of

6 4 2 0 2 0.1 0 Horizontal Vertical

Impact

Balance

0.1

0.2

Time (s) 0.3 End of Landing

Contact

Figure 1 Landing GRFs after a single-revolution initiated 0.6 m above ground level; data taken from McNitt-Gray et al. (2001). Peak GRFs would be more than twice as large after an uneven bar dismount beginning from 2.4 m.

possible landing states. The same impulse is required to stop a mass travelling at a given speed regardless of whether the mass is modelled as rigid or wobbling and regardless of the landing surface. The complexity of a wobbling mass model is not necessary for the present application because only the impulses required to stop the feet are calculated, not the GRF details or joint torques required to hold the body rigid. Dynamic balance was predicted by Pai & Patton (1997) by modelling the human as an inverted pendulum with a foot. Sets of CM positions and angular velocities that can be balanced in the presence of environmental, anatomical and physiological constraints, called feasible stability ranges, were identified. Their balance predictions using the simplified model encompassed 99.8% of empirical trajectories even though subjects could move all body segments to regain balance (Pai & Patton, 1997; Patton, Pai & Lee, 1999). Experimental measurements of landings that include impact and balance phases have identified common balanced landing characteristics. Drop landing experiments have measured muscle activation patterns, joint kinematics and GRFs after falls from various heights (Arampatzis et al., 2003; McNitt-Gray, et al., 1993; Seegmiller and McCaw, 2003) and onto different surfaces (McNitt-Gray et al., 1993). While strategies for different landing conditions were determined, the drop landing results are not completely applicable for gymnasts with larger pre-impact angular velocities. In a more relevant study, McNitt-Gray et al. (2001) quantified differences in gymnast multi-joint control strategies for three landings: following a singleforward somersault, backward somersault, and drop landing. Successful landing positions and orientations as a function of CM velocity were identified, but were limited to the subjects preferred landing strategies. Furthermore, it is unlikely that the subjects would have voluntarily approached balance boundaries after these relatively simple single-somersault manoeuvres. While models of both impact and balance have been developed previously, they have not been combined to identify balanceable sets of pre-impact landing positions and angular velocities following airborne movements with non-zero angular velocities. This can be done without the details of the peak, time to peak and duration of the GRF, by using impulses.

Reaction force (BW)

210

Sports Engineering (2007) 10, 209220 2007 isea

A.L. Sheets

Dynamic approximation of balanced landings

The goal of this paper is to create approximate models of each landing phase and combine them to identify balanceable landings following single- and doublebackward somersault uneven bar dismounts. The dismounts were chosen to investigate the influence of flight angular momentum on balanceable pre-impact landing conditions. Gymnasts can use this information to develop strategies that allow them to land without moving the feet.

CM +0 0 i

CM i

Methods
An algebraic impact model and a dynamic balance model approximate the two phases of a balanced landing after an uneven parallel bar dismount. A rigid-body inverted pendulum that rotates about a foot, similar that used in the dynamic balance model of Pai & Patton (1997) and Patton & Pai (1999), approximates the gymnast during impact and balance phases (Figure 3, Table 1). A dynamic balance model is used in this paper because of the potential for nonnegligible post-impact CM velocities. For example, with the CM between the toe and heel, after impact a static model neglecting a sufficiently large velocity away from the base of support would incorrectly conclude that the gymnast can balance. Conversely, if the CM is not above the feet after impact but has velocity toward the base of support, balance may be possible even though it is not predicted by a static model (Pai & Patton, 1997; Hof et al., 2005). Although there are different types of uneven bar dismounts, single- and double-backward somersault dismounts were chosen for this study because they are commonly performed and allow for investigation of numerous pre-impact angular velocities. The results may also interest coaches because high level gymnasts typically perform double somersault dismounts (with or without twists), while average gymnasts perform single somersault dismounts. Sign conventions for lean angle 0 and rotation rate 0 are positive when leaning and rotating backward, respectively, and are shown at initial ground contact (Fig. 2a). In flight the gymnast is assumed to rotate backwards (0 > 0) while translating forward (vx0 > 0). The body is assumed to be rigid at contact, and remains rigid while non-instantaneous vertical (y ) and horizontal (x) impulses (Fig. 2a) are applied to

x y a Fh = mg Ft = mg b

Figure 2 a Gymnast initial conditions (IC) 0, 0 for the impact model (typically 0 > 0 and 0 < 0) b ICs for the balance model. The terminal conditions after impact i, i are the ICs for the balance phase. Balance is achieved if fmin < f < fmax when f = 0 or if the CM stops above the foot between the toe and heel.

stop the feet (McNitt-Gray et al., 2001). Although not exact, the rigid body assumption may be an adequate approximation because the model requires the calculation of only total impulses and not the time duration of contact GRF and joint force; 0.04 s is too short for active feedback neural control and therefore muscles must be pre-activated. The lower extremity must be stiff to resist collapse during impact because there is a large pre-impact velocity (Arampatzis et. al., 2003). Even though body configuration (shape) during impact is assumed to be constant, impact duration and angular velocity are large enough that total rigid-body rotation (~0.3 rad) is non-negligible. Whole-body rotation is included by assuming that post-impact rotation occurs about the ankle and that angular velocity decreases linearly during impact. The assumptions are necessary to calculate impulse moment arms used in the non-instantaneous linear and angular impulse and momentum relationships approximating impact: y = m(vyi vy0) + mgti x = m(vxi vx0) Icm(i 0) = r(x cos avg y sin avg) (1) (2) (3)

2007 isea

Sports Engineering (2007) 10, 209220

211

Dynamic approximation of balanced landings

A.L. Sheets

where m is gymnast mass, g gravity, r the distance from ankle to CM, Icm the whole body moment of inertia about the CM, avg = 0 + (0 + i)t/4 the average angle between the line from ankle to CM and vertical, and subscripts 0 and i denote immediately before and after impact, respectively (Fig. 1). The inputs to the impulsemomentum eqns. 13 are pre-impact angle 0, measured between the line from ankle to CM and vertical, angular velocity 0, and horizontal and vertical CM velocities vx0 and vy0, respectively (experiments used to determine Icm, Hcm, vx0 and vy0 are described below). Eqns. 13 are solved for the impulses required to stop the feet, x and y, and the post-impact angular velocity, i. Assuming that the feet stop implies that vxi = i r cos i and vyi = i r sin i where i = 0 + (0 + i)t/2. Impact may cause i to vanish completely with the CM between the toe and heel, allowing standing without active balance. However, gymnasts can balance over a wider range of impact configurations and velocities using time-varying muscular torques. Post-impact balance generally involves all joints, but a conservative approximate balance model allows attenuation of i using only ankle moments due to a constant GRF (~mg, see Fig. 1) applied between the toe and heel (Fig. 2b). The body is assumed to be rigid, so only the limited torque exerted about the ankle could accelerate the CM and contribute to the GRF deviating from 1 BW. Assuming that balance is only possible if the CM stops (f = 0) above the feet, between the toe and heel, at a final angle f using the limited ankle moments, the post-balancing static boundaries can be calculated: f min = sin1 (dt/r) and f max = sin1 (dh/r). The static balance boundaries are roughly equal to the foot length divided by the ankle to CM distance and are not unique because r varies with the moment of inertia required to attain each angular velocity, 0, while conserving release Hcm. Assuming that an ankle torque limited to a GRF of one BW applied at the toe and heel is used to balance, it is possible to calculate minimum and maximum angles i min and i max, respectively, from which balance can occur. The given post-impact angular velocity i is used as one of the ICs in each of the balance differential equations: Iamin = mgr sin min + mgdt
..

Iamax = mgr sin max + mgdh

..

(5)

where dt and dh are horizontal distances from ankle to toe and heel, respectively, and Ia is the moment of inertia about the ankle (Table 1). Eqns. 45 are used to solve for the other unknown limiting ICs i min and i max that allow the CM to stop over the feet using maximum control at toe or heel, respectively. The range of balanceable post-impact angles i min< i < i max is thus determined for a given i. But the real question of interest is: for a given preimpact 0 what is the landing window or range of balanceable pre-impact angles 0min < 0 < 0max? The landing window is calculated for all possible 0min < 0 < 0max where 0min = Hcm/Imax and 0max = Hcm/Imin (Imin and Imax will be discussed in Fig. 3) using the following test algorithm for each candidate point [0, 0]: 1 Choose 0 such that 0min < 0 < 0max. 2 Choose an arbitrary 0 as a possible balanceable landing angle. 3 Use impact eqns. 13 to calculate i and i. 4 Use i and i as ICs for the balance eqns. 45 to calculate f when f = 0. 5 If fmin < f < fmax then 0 was balanceable, and [0, 0] is a feasible landing state. For a given 0, the value of 0 was gradually varied until f = fmin (and consequently 0 = 0min) and then increased until f = fmax (and 0 = 0max). By calculating the landing window for all possible 0 for a gymnast with a constant angular momentum Hcm (Table 1), a band of landing windows or a balanceable region in the 0, 0 space is identified. Gymnasts control angular velocity by varying Icm. Although the ranges {0min, 0max} above encompass all 0 at which landings can occur with the given Hcm and controlled Icm, it was not verified that a given 0 was achievable for the corresponding 0. Nevertheless, the gymnast should be able to access most, if not all 0 given in the results because Icm can vary (6.63 < Icm < 13.05 kg m2) over short periods (~0.2 s) compared to the longer flight time (~1 s). If 0 < 0min, the gymnast cannot create a large enough moment using the GRF at the toe and will fall forward, whereas if 0 > 0max the gymnast falls backward. When contact conditions [0, 0] lie within

(4)

212

Sports Engineering (2007) 10, 209220 2007 isea

A.L. Sheets

Dynamic approximation of balanced landings

the band of landing windows, the model predicts balance is possible using only ankle torques. Understanding the relationship between 0 and 0 is necessary to determine general balancing strategies. The impact and balance model body segment parameters and initial conditions were measured using a female collegiate gymnast who had provided informed consent. Inertial parameters were calculated using 95 measurements (Yeadon, 1990, Table 1). Dismount preimpact conditions were experimentally determined using motion analysis techniques for two single- and two double-backward somersault dismounts with the body fully straight. The arms remained above the head during the single-somersault dismount and the dropped to the sides after release during the double. Angular momentum (Hcm) and horizontal and vertical CM velocity (vxr and vyr, respectively) at bar release for each trial were calculated and averaged for the single- and double-somersault performances. Hcm and vxr were assumed to remain constant during flight, and impact conditions were calculated from release conditions (Table 2) using vx0 = vxr, vy0 = vyr gt. Although the experimental gymnast maintained a straight body position during dismount flight, a range of simulated body configurations was investigated
Table 1 Gymnast size and mass properties Height (m) 1.71 Mass (kg) 72.5227

dt (m)
0.19

dh (m)
0.05

Imin (kg m2)


6.63

Imax (kg m2)


13.05

(from straight to tucked) because the effect of preimpact angular velocities on balanceable pre-impact configurations is of interest in this paper. Even though the simulated gymnast impacts with numerous body configurations, an experimentally determined constant flight Hcm is assumed for the one- and twosomersault dismounts. With constant flight Hcm the gymnast must change her Icm to achieve different 0, but as Icm changes so does the radius from ankle to the mass centre, r. To calculate r as a function of the Icm at landing (Fig. 3), a flight simulation was used. The body configuration was changed from straight with the arms above the head (Imax) to maximally tucked (Imin) while rotating with the measured double somersault Hcm of 123.67 kg m2 s1. Hips, knees, and shoulders were flexed using maximal exertion, assuming a nominal set of joint torques (Cheng & Hubbard, 2005). The torques were calculated by multiplying a maximum isometric joint torque by functions of joint angle, angular velocity and exertion, and the amount of tuck was limited by gymnast strength. Mass centre location and Icm were calculated at discrete times as body configuration changed to create the relationship between r and Icm shown in Fig. 3. Although tighter tucks are possible during a single somersault, the same relationship was used for both dismounts because landings in tightly tucked positions are unlikely.

0.90

Radius, r, from ankle to CM (m)

Table 2 Experimentally measured gymnast release and pre-impact angular momentum and CM velocities

Imax 0.85 0.80 0.75 rmax 0.70 rmin 0.65 Imin 6 8 10 12 14

Hcm (kg m2 s1)


Single-revolution Double-revolution 69.52 123.67

vx 0 (m s1) 1.22 2.44

Bar release vyr (m s1) 4.47 3.87

Pre-impact vy 0 (m s1) 6.88 6.51

Abbreviations: dt and dh are horizontal distances from ankle to toe and heel, respectively. Imax is the gymnasts maximum moment of inertia about the mass centre (CM) when the body was straight with the arms above the head. Imin is the minimum moment of inertia about the CM when the gymnast is maximally tucked while rotating with the measured double-revolution Hcm of 123.67 kg m2 s1. Hcm is the flight angular momentum. vxr and vyr are horizontal and vertical CM velocity at bar release, respectively. vx0 and vy0 are pre-impact horizontal and vertical CM velocity, respectively.

Moment of inertia about the CM (kg m2)

Figure 3 Radius from ankle to CM as a function of the moment of inertia of a typical female collegiate gymnast

2007 isea

Sports Engineering (2007) 10, 209220

213

Dynamic approximation of balanced landings

A.L. Sheets

Results
Following a single- or double-backward somersault dismount, to land balanced the gymnast must increase pre-impact forward lean, 0, as the pre-impact angular velocity, 0, increases (Fig. 4). Even though forward lean increases with 0, the CM must be between the toe and heel at impact in order to balance after one somersault for all 0. This means that the pre-impact balance boundaries ([0min, 0], and [0max, 0]) lie inside the static landing window, the boundaries of balanceable configurations if the angular velocity were zero (Fig. 4). For the double-somersault dismount, the gymnast must also land with her CM
20

inside the static balance boundaries for all but the largest 0 (Fig. 4). For the gymnast to balance with these large values of 0, the CM must be slightly in front of the toe, or outside the static balance window. During impact the body continues to rotate about the ankle ( = i 0) and large 0 are attenuated, while smaller orientation and velocity changes occur during the balance phase. Body orientation changes ( = i 0) during impact increase with 0, causing the gymnast to contact the ground with more forward lean (a larger | 0|) to end balanced after a single- or double-somersault dismount (Fig. 5). For almost all successfully

15

double revolution dynamic balance landing region

double revolution static balance landing region

0 (rad s1)

0 min fall forward boundaries 10

0 max fall backward boundaries

single revolution dynamic balance landing region

single revolution static balance landing region

5 0.4

0.35

0.3

0.25

0.2

0.15 0 (rad)

0.1

0.05

0.05

0.1

Figure 4 Bands of dynamic balanceable landing windows, sets of [0, 0], and static balance windows following single- and doublebackward revolution dismounts

214

Sports Engineering (2007) 10, 209220 2007 isea

A.L. Sheets

Dynamic approximation of balanced landings

balanced landings, the CM begins in front of the ankle and rotates past the ankle during impact (yielding positive i). During the 0.04 s impact, the angular velocity slows significantly and can change direction if y and x that stop the feet are large enough. After a single- or double-backward somersault dismount, there is forward rotation at the end of impact (i < 0) if ground contact occurs on the fall forward boundary (0min, Figure 6); therefore, after impact a torque must be produced using the GRF at the toe to not fall forward. If a single-somersault dismount is landed in the maximum balanceable 0, for most 0 a small i results
0.14

in the same direction as during flight (Fig. 6). Balance would be attained by exerting a torque created by a GRF at the heel. Most pre-impact 0 result in forward rotation (i < 0) after impact because a larger torque can be created with the toe than with the heel. Larger pre-impact CM horizontal and vertical velocities, vx0 and vy0, require larger linear impulses to stop the feet (Fig. 7). Although the double-somersault dismount has a slightly smaller vy0, its vx0 is twice as large as that of the single-somersault dismount, causing y to be smaller and x to be much larger, respectively. Also, |y| increases only slightly as |0| increases for both dismounts, although x varies more.

0.12

0 max fall backward boundary

0.1 0 min 0.08 0 max boundary 0.06 0 max Double revolution

i (rad)

0.04

0.02 Single revolution 0 0 max 0.02 0 min fall forward boundary

0 min

Single Double 0 min boundary

0.04

0.06 0.4

0.35

0.3

0.25

0.2 0 (rad)

0.15

0.1

0.05

Figure 5 Regions of post-impact body position, i, resulting from impact in a given pre-impact orientation, 0, following a single- and double- backward revolution dismount

2007 isea

Sports Engineering (2007) 10, 209220

215

Dynamic approximation of balanced landings

A.L. Sheets

0.2 0 max fall backward boundaries Post-impact backward rotation 0 Post-impact forward rotation Single revolution

0.2

0.4 Double revolution


i (rad s1)

0 max 0.6

0 min

0.8

0 min fall forward boundaries

1.2 0.4

0.35

0.3

0.25

0.2 0 (rad)

0.15

0.1

0.05

Figure 6 Post-impact angular velocity, i, resulting from impact at a given 0 following single- and double-backward revolution dismounts

40 60

0 min fall forward boundary 0 max

530 525 520 515 510 505 500 0.4 Double revolution vyo = 6.51 m s1 Single revolution vyo = 6.89 m s1

x-direction impulse, x (N s)

80 100 120 140 160 180 0.4 0 max

0 max fall backward boundary 0 min fall forward boundary Double revolution vxo = 2.44 m s1 0 min

0 max fall backward boundary 0.3 0.2 0 (rad) 0.1 0

y-direction impulse, y (N s)

Single revolution 0 min 1 vx0 = 1.22 m s

0.3

0.2 0 (rad)

0.1

Figure 7 X- and y- direction impulses required to stop the feet following single- and double-backward revolution dismounts

216

Sports Engineering (2007) 10, 209220 2007 isea

A.L. Sheets

Dynamic approximation of balanced landings

The landing window width (range of balanceable 0, 0max0min) for a single- and double-somersault dismount increases with angular velocity (Fig. 8). Although the landing window is slightly larger at large values of 0, the gymnast is in controllable landing orientations longer when 0 is smaller (Fig. 9).
0.105 * 0.100 * 0.095 0.090 0.085 0.080 0.075 5 ^ 10 0 (rad s1) 15 ^ * Imin, 0 max ^ Imax, 0min 20 Double revolution

0 max 0 min (rad)

Single revolution

Figure 8 The range of balanceable pre-impact landing orientations increases as pre-impact angular velocity increases following single- and double- backward revolution dismounts
0.015 0.014 0.013 0.011 Single revolution ^

* Imin, 0 max ^ Imax, 0min

Time (s)

0.01 0.009 0.008 0.007 0.006 0.005 5 * 10 0 (rad s1) 15 20 ^ * Double revolution

Figure 9 Time in balanceable 0 range decreases as 0 increases following single- and double-backward revolution dismounts

Discussion
The previous results indicate that a two-part model is needed to investigate gymnastics landings because large changes occur in and during impact and

post-impact angular velocities i need not be exactly zero, but rather only small enough to attenuate thereafter using ankle joint torques. General landing strategies for a single- and double-backward somersault dismount were calculated using this model. To balance after either dismount, a gymnast must impact the ground with the CM between the toe and heel for almost all 0 (Fig. 4). This finding is different from previous experimental results (McNitt-Gray et al., 2001) of successfully balanced landings following backward or forward single-somersaults with lower angular velocities, and initiated from a lower height (0.6 m instead of 2.4 m). This is because in the previous study the gymnast either rotated backward and travelled backward or rotated and travelled forward, while in this simulation the gymnast rotated backward and travelled forward. In the McNitt-Gray study, vx0 was in the opposite direction, x creates an angular impulse about the CM to increase the whole body angular velocity (eqn. 3), and the contribution of y to the angular impulse was smaller than in a dismount because vy0 was only half as large. With larger 0 and smaller Icm, a gymnast must impact with a larger angle relative to vertical |0| than one with larger Icm and smaller 0. This is primarily because changes in body orientation during impact increase with 0, because is assumed to slow linearly; so there must be more forward lean to end in a balanceable configuration after rotating farther. Vertical and horizontal impulses, y > 0 and x < 0, create large negative angular impulses about the CM, opposing the positive Hcm, and can result in either a positive or negative i (Fig. 2a). When the gymnast contacts the ground inside the band of landing windows with [0, 0] closer to the fall-forward than the fall-backward boundary (Fig. 4), the impact impulses cause to change direction, resulting in a negative i and thereafter balance occurs using the GRF at the toe (Figs. 2, 4 and 6). However, for landings closer to the fall-backward boundary following a single somersault, the angular impulses are too small to result in positive i and the gymnast must balance with the GRF at the heel (Figs. 2, 4 and 6). Most balanceable [0, 0] result in forward lean (0) and/or rotation (0) because a larger moment can be created using the GRF at the toe than at the heel due to the longer moment arm (Figs. 5 and 6).

2007 isea

Sports Engineering (2007) 10, 209220

217

Dynamic approximation of balanced landings

A.L. Sheets

The double-somersault dismount requires a larger flight Hcm than a single-somersault dismount (Table 1), and during impact Hcm is attenuated by the contributions of the linear impulse to the angular impulse. The value of vx0, and consequently the value of x required to stop the feet, is approximately twice as large for the double (Table 2), while vy0 and y are similar for the two dismounts (Fig. 7). The y following the single somersault was slightly larger than with the double because the flight time was slightly longer: 1.06 s and 1.16 s, respectively. Following both dismounts for landing orientations near vertical, the value of x contributes slightly more to the angular impulse than y, even though y is between two and ten times larger, because the impulse moment arm is much longer (eqn. 3). Although y does not vary much with changes in 0, it contributes more when the gymnast is leaning farther forward at impact, because small increases in |0| cause roughly proportional increases in the horizontal y moment arm. The total angular impulse created by the linear impulses to slow Hcm is also affected by changes in 0 because impulse moment arm length r varies with values of Icm. These results could be used to improve gymnast landing consistency by determining the single balanceable [0, 0] that provides the largest margin of error. Although as 0 increases there are more values of 0 that are balanceable for a given 0 (Fig. 8), the gymnast is in controllable landing orientations longer when rotating more slowly (Fig. 9) and as a result may be more likely to end balanced. A similar definition for robustness in gymnastics was introduced by Hiley & Yeadon (2003). Also, the gymnast is appropriately oriented following a single-somersault dismount for longer than after a double-somersault dismount, for all possible pre-impact angular velocities. It is probable that the most robust landing strategy would include consideration of both how close the point [0, 0] is to a fall boundary and the value of 0. Although conclusions can be drawn from the present study, future models should evaluate the validity of the assumptions that the gymnast: 1 2 3 4 5 is able to remain rigid during impact rotates only about the ankle during balance applies a constant GRF GRF is limited to 1 BW applies the GRF at the end of the toe or heel.

Assumptions 24 underestimate the gymnasts control capabilities; therefore it is probable that the presented successful landing windows slightly underestimate the size of the actual [0, 0] regions. Because impact time is so short, the assumption that the gymnast does not actively control joint angle changes is probably good. The muscles must be preactivated before impact because the 0.04 s time of impact is much shorter than human reaction time. Although some joint motion is likely during impact, the rigid body assumption may be adequate for understanding the relationships between 0, 0, vx0, and vy0 during impact and balance. If the joints did move, the effect would be to shorten the radius from the ankle to the CM, thus decreasing the linear impulse contributions to angular impulse, increasing angular velocity. This would allow for balance after contacting the ground with the CM farther forward than predicted (0 < 0min), but may also cause a fall backward if contact occurs at the 0max boundary. Joint motion during impact could also increase the balanceable [0, 0] combinations due to moments created by limb acceleration. It is possible that the single-segment dynamic balance model produces realistic results because the slightly overestimated length of the functional base of support increases the landing window size, and thus compensates for the conservative ankle torque limitations and simplified joint control strategy. In reality, balance involves multiple joints and the functional base of support is shorter than the foot, and varies according to age and a number of unknown factors (Patton et al., 1999). Iqbal & Pai (2000) found that increasing balance model complexity to include control at the hip, knee and ankle joints expanded the balanceable region, most notably for small angular velocities and initial CM positions farther outside the base of support in the direction opposite CM velocity. It is likely that the balanceable region would also expand if a more complex balance model were used in this study. By combining approximate impact and balance models, other gymnastics landings with different preimpact angular and linear velocities may be investigated (eg. forward rotation and forward translation or backward rotation and backward translation). Alternatively, more complex models that remove some of the present models limiting assumptions may find larger sets of balanceable landing regions.

218

Sports Engineering (2007) 10, 209220 2007 isea

A.L. Sheets

Dynamic approximation of balanced landings

References
Arampatzis, A., Morey-Klapsing, G., & Bruggemann, G. P. (2003) The effect of falling height on muscle activity and foot motion during landings. Journal of Electromyography & Kinesiology, 13, 533544. Cheng, K.B. & Hubbard, M. (2005) Optimal compliantsurface jumping: a multi-segment model of springboard standing jumps. Journal of Biomechanics, 38, 18221829. Gittoes, M.J.R., Brewin, M.A. & Kerwin, D.G. (2006) Soft tissue contributions to impact forces simulated using a four-segment wobbling mass model of forefoot-heel landings. Human Movement Science, 25, 775787. Gruber, K., Ruder, H., Denoth, J. & Schneider, K. (1998) A comparative study of impact dynamics: wobbling mass model versus rigid body models. Journal of Biomechanics, 31, 439444. Hiley, M.J. & Yeadon, M.R. (2003) The margin for error when releasing the high bar for dismounts. Journal of Biomechanics, 36, 313319. Hof, A.L., Gazendam, M.G.J. & Sinke, W.E. (2005) The condition for dynamic stability. Journal of Biomechanics, 38, 18. Iqbal, K. & Pai, Y.-C. (2000) Predicted region for balance recovery: motion at the knee joint can improve termination of forward movement. Journal of Biomechanics, 33, 16191627. McNitt-Gray, J.L. (1991) Kinematics and impulse characteristics of drop landings from three heights. International Journal of Sports Biomechanics, 7, 201204. McNitt-Gray, J. L., T. Yokoi & Millward, C. (1993) Landing strategy adjustments made by female gymnasts in response to drop height and mat composition. Journal of Applied Biomechanics, 9, 173190.

McNitt-Gray, J.L., Hester, D.M.E., Mathiyakom, W. & Munkasy, B.A. (2001) Mechanical demand and multijoint control during landing depend on orientation of the body segments relative to the reaction force. Journal of Biomechanics, 34, 14711482. Nigg, B.M. & Liu, W. (1999) The effect of muscle stiffness and damping on simulated impact force peaks during running. Journal of Biomechanics, 32, 849856. Pai, Y.-C. & Patton, J.L. (1997) Center of mass velocityposition predictions for balance control. Journal of Biomechanics, 30, 347354. Pain, M.T.G. & Challis, J.H. (2002) Soft tissue motion during impacts: Their potential contributions to energy dissipation. Journal of Applied Biomechanics, 18(3), 231242. Pain, M.T.G. & Challis, J.H. (2006) The influence of soft tissue movement on ground reaction forces, joint torques and joint reaction forces in drop landings. Journal of Biomechanics, 39, 119124. Patton, J.L., Pai, Y.-C. & Lee, W.A. (1999) Evaluation of a model that determines the stability limits of dynamic balance. Gait & Posture, 9, 3849. Seegmiller, J. G. & McCaw, S. T. (2003) Ground reaction forces among gymnasts and recreational athletes in drop landings. Journal of Athletic Training, 38, 311314. Yeadon, M.R. (1990) The simulation of aerial movement. II. A mathematical inertia model of the human body. Journal of Biomechanics, 23, 6774.

2007 isea

Sports Engineering (2007) 10, 209220

219

Das könnte Ihnen auch gefallen