JOURNAL OF THEORETICAL AND APPLIED MECHANICS 42, 4, pp. 927-944, Warsaw 2004 BIOMECHANICAL MODELLING FOR WHOLE BODY MOTION USING NATURAL COORDINATES Adam Czaplicki Academy of Physical Education in Warsaw, External Faculty of Physical Education in Biała Podlaska, Department of Biomechanics, Poland; e-mail: czaplicki@poczta.onet.pl Miguel T. Silva Jorge C. Ambrósio IDMEC – Institute of Mechanical Engineering, Technical University of Lisbon, Portugal e-mail: pcms@mail.ist.utl.pt; jorge@dem.ist.utl.pt The study of spatial human movements requires the development and use of a three-dimensional model. The model proposed here has 44 degrees-of- freedom and it is described using natural coordinates, which do not require an explicit definition of rotation coordinates. The biomechanical model con- sists of 16 anatomical segments composed of 33 rigid bodies. Joint actuators are introduced into equations of motion of the multibodymodel bymeans of kinematic driver constraints in order to reflect the effect of the muscle forces about each anatomical joint. After associating a Lagrangemultiplier to each joint actuator, the torques that representmuscle forces become coupled with the biomechanicalmodel through the Jacobianmatrix of the underlyingmul- tibody system. The developed model is applied to identify net torques and reaction forces at the anatomical joints in application cases that include the take-off to aerial trajectories and standing backwards somersault. Key words: biomechanical model, multibody dynamics, inverse dynamics, internal forces 1. Introduction The natural coordinates were first proposed by Jalón et al. (1986) and descri- bed later indetail byJalónandBayo (1994).Theyprovideanatural framework for the description of complete spatial motion of multibody systems without using rotation coordinates explicitly. Not only these coordinates lead to very 928 A.Czaplicki et al. efficient numerical models but also there is a direct connection between the coordinates defining the bodies and the anatomical landmarks associated to kinematic data acquisition typical to biomechanical modelling. Despite the fact that the natural coordinates are commonly used to investigate different types of mechanical systems, they are sporadically applied to biomechanics as proposed by Celigüeta (1996), Silva et al. (1997), Silva and Ambrósio (2002), and Ambrósio et al. (2001). Therefore, one of the purposes of this work is to demonstrate that the modelling in biomechanics by means of natural co- ordinates is not only efficient and attractive but also a natural way to map the body anatomical landmarks into coordinates defining the biomechanical system. The number of natural coordinates used to represent a biomechanical mo- del is always larger than its number of degrees-of-freedom, which means that they are related by proper algebraic constraint equations. These are usually nonlinear and play amajor role in the kinematic anddynamic analysis ofmul- tibody systems. The number of independent algebraic constraint equations is equal to the difference between the number of natural coordinates and the number of degrees-of-freedom of a system. There are some aspects in which a set of natural coordinates is advantageous. First, the wholemultibody system is defined directly in the global reference frame. Second, natural coordinates are usually defined at the joints of the system and shared by adjacent bo- dies, which reduces the total number of coordinates required to represent the model and considerably simplifies the definition of the joint constraint equ- ations. Third, with natural coordinates the constraint equations, that arise from rigid body and joint conditions, are linear or quadratic. Thismeans that their contribution to the Jacobian matrix of the constraints is a constant or linear function of the natural coordinates, which leads to efficient numerical implementations. The aim of this paper is to present a general purposemethodology for the three-dimensional modelling in biomechanics. The approach is based on natu- ral coordinates, and it focuses on the identification of internal reaction forces and net torques at anatomical joints of an individual performing selectedmo- tor tasks. Twodifferent humanactivities have been chosen to demonstrate the method offered: the take-off for a jump and the standing backwards somer- sault. The selected jump is being very often preformed when passing natural obstacles like puddles or ditches, and it is undoubtedly a frequent human ac- tivity. Many researches have so far conducted analyses of long jumps (e.g., Hay, 1993; Lees et al., 1994) and of other relatedmotor task like vertical jum- ping (e.g., Pandy et al., 1990, 1992; Bobbert and Schenau, 1988; Selbie and Biomechanical modelling for whole body motion... 929 Caldwell, 1996; Eberhard et al., 1999). However, in the case of the specified jump, these articles can be helpful in a very limited way since they are either descriptive by nature or the results reported concern two-dimensional models only. Numerous standing backwards somersault analyses deal mostly with the methodology of learning this stunt neglecting its mechanical aspects (e.g., Mieczkowski, 1982). The quantitative description of the standing backwards somersault is important for both cognitive and practical reasons, leading to better understandingof the athletemovements and formsabasis formore con- scious mastering of the evolution of the somersault technique. In the process of studying the standing backwards somersault in terms of the biomechanical analysis, a three-dimensional methodology is proposed in this work and is de- monstrated in amovement of the human body that leads to a change in space positions of anatomical segments by a complete revolution by 360 degrees, emphasizing the robustness of the methods adopted. 2. Biomechanical model The study presented here uses the biomechanical model shown in Fig.1a, which consists of 16 anatomical segments. The biomechanical model is descri- bed by natural coordinates that are used to construct its kinematic structure made of 33 rigid bodies branching from the pelvis in open chain linkages (Silva and Ambrósio, 2002). The rigid bodies that form the neck, arms, fo- rearms, thighs, shanks, upper torso (numbered in Fig.1b from 19 to 25) and the lower torso (numbers 6, 7, 8) are defined by the Cartesian coordinates of two points and one unit vector each. The hands, feet and head are defined by the coordinates of three points and one unit vector. The first type of the rigid body has a simpler kinematic structure but generates a non-constant mass matrix in the equations of motion of the system. The second type of the rigid body has a more complex kinematic structure but generates a constant mass matrix (Jalón and Bayo, 1994). The complete set of rigid bodies is described by means of 25 points and 22 unit vectors, accounting a total number of 141 natural coordinates. The most common way of modelling the joints in the natural coordina- tes environment, corresponding to the sharing points and vectors among the adjacent segments, is depicted in Fig.2. The kinematic structure of the biomechanical model, shown in Fig.1c, is worth noticing as it suggests a more detailed discussion. The calculation of the internal joint reactions is usually very important from the biomechanical 930 A.Czaplicki et al. Fig. 1. Biomechanical model. (a) Representation of sixteen anatomical segments; (b) the structure of thirty three rigid bodies (with body numbers inside circles) used to represent the sixteen anatomical segments, and fifteen kinematic joints (with the joint numbers in bold); (c) an exploded view of the kinematic structure of the biomechanical model with indication of the points (represented in italic) and unit vectors Fig. 2. The model of anatomical joints with natural coordinates: (a) the spherical joint for the hip; (b) the revolute joint for the elbow; (c) the universal joint for the ankle standpoint. To calculate their anterior- posterior, medial-lateral and vertical components, all thekinematic joint constraintsmustbe explicitly defined.This is achieved by introducing additional rigid bodies, points and unit vectors at the joints that are physically the same but mathematically distinguishable (Silva, 2003). As a result, the kinematic structure of the biomechanical model Biomechanical modelling for whole body motion... 931 becomes more complex and the biomechanical model is provided with revo- lute and universal joints only, as represented in Fig.1c. The universal joints are located in the ankle, in the radioulnar articulations, between the 12th tho- racic and 1st lumbar vertebrae (designated by lower-upper torso joint) and between the 7th cervical and 1st thoracic vertebrae (referred to as the upper torso-neck joint). Sincemost of the joints are defined naturally by the sharing points and vectors between the contiguous segments, there is a total number of 97 non-redundant kinematic constraints only. The model has 44 degrees of freedom that correspond to 38 rotations about the revolute and universal joints, and 6 degrees-of-freedom associated with the pelvis, which is treated as the base body. The degrees-of-freedom of the biomechanical model proposed are numbered in Fig.3a. Fig. 3. (a) Degrees of freedom of the biomechanical model; (b) the joint actuator driving the knee joint in the sagittal plane One of the forms of calculating internal net torques at the joints of the biomechanical model is to assume that each degree-of-freedom is driven by an adequate joint actuator, such as the one depicted in Fig.3b. Joint actuators are introduced into equations of motion of the biomechanical model bymeans of kinematic driver constraints of the scalar product type, which are described in Table 1. The body segment parameters for the biomechanical model, which include the body mass, principal moments of inertia, segment lengths, and location of the center of mass of each segment with respect to the proximal joint, are estimated from the data reported by Laananen et al. (1983), Laananen (1991) 932 A.Czaplicki et al. andZatsiorsky (1998).All theseparameters are scaled according to the subject height andmass using the procedures proposed by Laananen (1991). 3. Multibody formulation Individual rigid bodies of the biomechanicalmodel are describedbymeans of a set of natural coordinates that is composed of theCartesian coordinates of several points and unit vectors located at the joints and segment extremities. The vector of generalized coordinates of the whole biomechanical model is denoted by q= [q1,q2, . . . ,qn] > (3.1) where n=3(np+nv) is the total number of natural coordinates and np and nv are the total number of points and unit vectors of the model, respecti- vely. The number of natural coordinates is always higher than the number of degrees-of-freedomof the system,whichmeans that somekinematic constraint equations need to be added to unequivocally define the kinematic structure and topology of the biomechanical model. In addition to the joint and driving constraints, there are also rigid body constraints utilized in the multibody approach presented. They express physical properties of rigid bodies such as constant distance between two points, constant angle between two segments or constant length of a unit vector. The latter condition implies the norma- lization constraint a2x + a 2 y + a 2 z = 1 on the coordinates of the unit vector a= [ax,ay,az] >. The physical interpretation of a Lagrangemultiplier associa- ted with this constraint when performing the inverse dynamic analysis is an axial pair of forces applied at the extremities of the unit vector. The rigidbody and driving constraints may be represented by the scalar product equation Φ sp(q, t)=v>u−LvLucos(〈v,u〉(t)) =0 (3.2) where v and u are two generic vectors used in the definition of rigid bodies, Lv and Lu are their respective norms and 〈v,u〉(t) is the angle between them, which may be time dependent. It is worth pointing out that the parameters (Lu,Lv and 〈v,u〉(t)) in Eq. (3.2) describe constraints imposed on u and v. They have constant or varying values (for driving constraints) and are com- puted (for non-driving constraints) only once at the beginning of the analysis in the pre-processing phase. Equation (3.2) presents some innovative aspects when compared with the original formulation proposed by Jálon and Bayo (1994), since it comprises in a single analytical expression several kinematic Biomechanical modelling for whole body motion... 933 constraints with different physical meanings, depending on the process of cal- culation of the vectors v and u. Considering that ri, rj, rk and rl are the Cartesian coordinates of the points i, j, k and l, and that a and b are unit vectors, the most relevant kinematic constraints involving the scalar product and their physical meanings are presented in Table 1. Table 1.Physical meanings of the scalar product constraints Constraint v u Lv Lu 〈v,u〉 Graphical description representation Constant distance between Bji Bji Lij Lij 0 points i and j Unit module vector a a 1 1 0 Constant angle between unit a a 1 1 α vectors a and b Constant angle between segment rij Bji a Lij 1 β and unit vector a Constant angle between segments Bji Bik Lij Lkl γ rij and rkl Rotational driver around revolute joint located in point i, Bji Bki Lij Lik φ using segments rij and rik where Bij =(ri−rj) and φ= f(t). With the natural coordinates, the constraint equations that arise from a rigid body and joint conditions are quadratic, so their Jacobian matrix is a linear function of the natural coordinates, which is very useful from the programming standpoint. In addition, Equation (3.2) may express rheonomic constraints when the angle 〈v,u〉 varies in time, which is used to define ro- tational driver actuators, such as the one presented in Fig.3b. All constraint equations are assembled in a single vector denoted as Φ(q, t)=0 (3.3) 934 A.Czaplicki et al. The biomechanical model is implemented in a general purpose multibody code (Nikravesh, 1988; Haug, 1989; Schiehlen, 1997; Silva, 2003). Thedynamic equations of motion for the model can be written in the generic form as Mq̈+Φ>q λ= g (3.4) where M is the global mass matrix of the system, Φq the Jacobian matrix of the constraints, q̈ the vector of generalized accelerations, g the genera- lized force vector and λ the vector of Lagrange multipliers. Note that the product Φ>q λ in Eq. (3.4) represents the joint reaction forces and the joint net moments. A unique vector λ is obtained when solving Eq. (3.4) during the inverse dynamics analysis. This is because the number of unknowns to cal- culate became equal to the number of equations of motion after introducing the driving constraints. It is worth pointing out that when solving the inverse dynamics problem one can obtain real values for the net torques (limited by the accuracy of the estimates of the angular accelerations only), but not real values for the reactions at the joints. Irrespective ofmethods applied, whether it is the Lagrange multipliers or conventional Newton-Euler approach, they do not take into account the contributions of individual muscle forces to the resultant joint reactions. Giving a graphic description of this fact, it is as if the biomechanical model was driven by torsional actuators located directly at the joints, instead of muscles. Nevertheless, the joint reactions computed this way are of great importance in the presence of transient external forces, and they can be easily adjusted after solving the individual muscle distribution problem at a particular joint. 4. Data acquisition An adult male with the age of 23, height of 168 cm and a body mass of 68 kg, carried out several jumps and back somersaults. Among different trials, one of each activities was chosen as the most representative and was used in this work to illustrate the methodologies proposed. In all jumps, the force-plate data and the body motion kinematic data were synchronized and recorded simultaneously. The ground reaction forces were measured using a Kistler 9281B force platform at a sampling frequency at 1000Hz, while the bodymotion was videotaped at 50Hz by 4 synchronized cameras. The global reference framewas attached to the center of the force platform.The positions of the 23 anatomical points were used to reconstruct themotion. It is illustra- ted by two pictures shown in Fig.4a, obtained during the beginning and the Biomechanical modelling for whole body motion... 935 end of the take-off phase. The coordinates of all anatomical points projected in the video frames, were digitized manually. The DLT technique was used to calculate the spatial coordinates of the anatomical points (Abdel-Aziz and Karara, 1971). The raw data consisting on the spatial coordinates of the ana- tomical points was smoothed by means of a low-pass Butterworth 2nd order filter with zero-phase lag. Proper cut-off frequencies for every segment of the biomechanicalmodelwere chosen after performinga residual analysis (Winter, 1990). Such a residual analysis provides, for the type ofmotion under analysis, cut-off frequencies that rangebetween 4 and6Hz.Taking into account the first investigated motor task, the average RMS differences (between raw and smo- othed data) of 6.5mm, 3.0mmand 3.3mmwere estimated for the x, y, and z coordinates, respectively. TheDLTprocedure did not ensure constant lengths of the anatomical segments during all instants of time. Therefore, a techni- que that ensures the kinematic consistency of themotion data was applied to avoid further problems in the estimation of the net torques and reactions at the joints (Silva and Ambrósio, 2002). In the cited work, the time histories of the net torques at the ankle, knee and hip joints during gait obtained for consistent as well as non-consistent kinematic data were presented. Fig. 4. Views from cameras 1 and 3; (a) long jump, white dots mark the positions of the 23 anatomical points; (b) standing backwards somersault, the end of take-off (left) and the beginning of landing (right) 5. Results 5.1. Long jump Thefirst step inorder toapply themethodsdescribed is to acquire the force and kinematic data according to the data acquisition procedures proposed. The time characteristics of the measured ground reaction forces are shown in 936 A.Czaplicki et al. Fig.5. This force data presents significant differences from that obtained in the gait analysis not only due to the shapes of curves but also because of their magnitude. The maximum vertical reaction is almost four times larger than its equivalent in the gait and the peak value of the medial-lateral component is seven times higher than its analogue in the gait (Silva, 2003). It is worth noticing that the peak value of the anterior-posterior component of the force is larger thanwhat is observed during jumps performed on a trampoline (Blajer and Czaplicki, 2003). An aliasing error during synchronization of the force plate data with the frequency of the video cameras is also visible. Fig. 5. Normalized vertical (Rz), anterior-posterior (Rx) andmedial-lateral (Ry) components of the ground reaction forcemeasured in the force platform. The markers denote the time instants for which there are video frames available. BW – body weight The inversedynamicanalysis of thebiomechanicalmodelwasfirstly carried out to find the net moment of forces in the anatomical joints of the lower extremities. The results obtained in the inverse dynamic analysis of the jump are presented in Fig.6. At the beginning of the trial, the jumper is in the air. Having the trunk in a forward-lean position, the jumper raises it to an upright position, which is achieved when the foot touches the ground. The net torque in the hip joint starts with positive values because the described movement requires the hamstrings and gluteus maximus to cooperate with each other. The large peak of the net torque in the hip joint, occurring at the beginning of the contact phase of the foot with the ground at t=0.08s, reflects the impact nature of the ground reaction force at that time. The net torque has a counterclockwise direction in the frontal plane, which means that the hip abductors, like glutei muscles and tensor fasciae latae, are active. The clockwise direction of the net torque in the sagittal plane in the hip joint together with the net torque oriented opposite in the knee joint indicate strong activity of vasti muscles. The ankle joint carries a considerable load during the support phase as a consequence of the stimulation of the triceps Biomechanical modelling for whole body motion... 937 surae muscle generating powerful ankle plantarflexion in order to push the body forward and to prevent the shank bending down to the ground. Finally, there is a distinguishable net torque in the hip joint in the transversal plane at the beginning of the airborne phase, when the jumper starts moving his thighs towards the trunk. The activity of the hip adductors, gluteusmaximus and iliopsoas muscles seems to ensure the medial rotation of the thigh. Fig. 6. Normalized net torques at the basic joints of the supporting leg of the jumper. BM – bodymass Fig. 7. Normalized reaction forces in the selected joints of the supporting leg during the jump The joint reaction forces, associated with the type of analysis carried out, are depicted in Fig.7. Note that using the joint torques to represent the mu- scle actions, the joint reaction forces are simply an approximation of the real forces, which can only be obtainedwith a completemusclemodel. It is clearly 938 A.Czaplicki et al. visible, when comparing the curves presented in Fig.5 and Fig.7, considera- ble damping of the vertical reaction force between the ground and hip joint that ranges from 4 to 2 times the bodyweight. The anterior-posterior and the medial- lateral components of the resultant reaction in the knee and in the hip joint have practically opposite directions during the take-off phase of the jump, which means that the femur presses the knee joint anteriorly and me- dially, being simultaneously pressed posteriorly and laterally at the hip joint. 5.2. Standing backwards somersault Having been an elite acrobat several years ago, the jumper performed the standing backwards somersault extremely well with a well developed techni- que, as represented inFig.4b. The upper and lower extremities were collateral during the stunt. The feet, having the same initial position in the anterior- posterior direction, were shifted only 0.8 cm apart from each other after lan- ding. Because the external forces were measured using one force plate only, the correct technical performance of the jump allows assuming a symmetrical distribution of the ground reactions for the feet. The application point of the ground reaction was established on the heel-metatarsal line coinciding with the cast of the center of mass of the body on the force plate in the anterior- posterior direction. The time characteristics of the measured ground reaction forces acquired are shown in Fig.8. The largest value, of about 10 times the body weight at landing, is reached by the vertical ground reaction force. A decrease in the force, below the bodyweight during the first 0.5s of the stunt due to a down- ward acceleration of the body performing a preparatory countermovement is also noticeable. This is followed by the overweight phase marked by a double humped pattern similar to that observed in high jumping (e.g., Pandy et al., 1992). Peakmagnitudes of the anterior-posterior component of the horizontal ground reaction lie in the vicinity of 3 times the bodyweight during landing. All presented characteristics distinctly emphasize the impact nature of the ground reaction during the beginning of the landing phase, and an aliasing error between the force plate and video data. The time characteristics of the net torque components for the joints of the right leg are depicted in Fig.9. The largest magnitudes of the net torques occur in the sagittal plane. Note that the My(t) characteristic in Fig.9 is split into two parts with different scales for the net torque to emphasize the characteristics of the lower leg torques during the take-off phase. The negative sign of the torque in the hip and ankle joint, in the second half of the support phase, indicates a common action of the quadriceps and triceps suraemuscles. Biomechanical modelling for whole body motion... 939 Fig. 8. Time-histories of the components of the ground reaction force measured in the force platform (left) and their expanded views during the beginning of landing (right). Themarkers denote the time instants for which there are video frames available Just prior to the airborne phase, the net torque in the hip starts changing its direction anticipating a later movement of the thighs towards the trunk. The shapes of all curves for the net torques during flight are similar to those reported in theworkbyBlajer andCzaplicki (2001). The landing phase begins with a considerable load of the hip joint, changing rapidly from 6Nm/BM (400Nm) to more then −10Nm/BM (−700Nm). Such an instantaneous rise in themagnitude of the net torque cannot be done by themuscles alone. This impact seems to be conveyed by joints and ligaments and followed almost immediately by a strong action of the hip extensors like gluteus maximus, adductormagnus, semimembranosus and semitendinosus, and knee extensors. The results obtainedduring landingare qualitatively similar to thosepresented 940 A.Czaplicki et al. in the work of McNitt-Gray et al. (2001), where the landing phase of several different jumps, including backwards somersault, was subjected to a detailed dynamical analysis in a two-dimensional space. Fig. 9. Normalized net torques at the anatomical joints of the right leg of the jumper Themagnitudes of the net torque in the frontal and transversal plane are one order lower to those achieved in the sagittal plane. A remarkable eversion in the ankle joint at thebeginningof landing isworthnoticing. Simultaneously, the feet and the shank are rotated medially as a consequence of the fact that the jumper tries to hold his position standing on his fingers, as illustrated in Fig.4b. Finally, there are also distinguishable net torque values in both planes during the secondhalf of the supportingphase,which suggests the involvement of passive elements of the joints in these directions. The results of identification of reaction forces in the selected joints in the case of the standing backwards somersault are shown in Fig.10. It is clearly visible, when comparing the curves presented in Fig.8 and Fig.10, the remar- kable damping of the vertical reaction between the ground and hip joint that ranges from 5 to 1.5 times the body weight. The shape of the vertical joint reaction during take-off and landing correspond closely to the shape of the ground reaction force. One should emphasize the positive signs of the verti- cal joint reaction at the beginning of the airborne phase and the considerable Biomechanical modelling for whole body motion... 941 values of the anterior-posterior reaction in the hip joint, changing suddenly from 0.7N/BW (466,̇) tomore than −1.1N/BW (−740N) during the impact phase of landing. Fig. 10. Normalized reaction forces in the selected joints of the right leg during the standing backwards somersault 6. Conclusions In thepresentworkageneral three-dimensionalmultibodymethodology for calculating the net torques and reactions in joints has beendemonstrated.The methodology consists of a biomechanical model, data acquisition techniques and an inverse dynamics analysis procedure integrated in a compact software package. The biomechanical model of the human body is constructed with rigid bodies interconnected by revolute and universal joints using the benefits of a general multibody formulation based on natural coordinates, well suited to define rigid bodies and kinematic joints and to a direct mapping between anatomical landmarks and body coordinates. The biomechanical model is controlled by joint and driver actuators. The Lagrange multipliers assigned to all driver actuators effectively represent in- ternal forces in the joints of the biomechanical model, either due to reaction forces or net moment of forces. The solution to the problem of inverse dyna- mics is unique and non-redundant, but it should be emphasized that the joint reactions obtained this way are underestimated. The application of themethodology developed to a long jump and a stan- dingbackwards somersault allows the identificationof the internal forcesacting in the joints of the biomechanical model of the human body. The results have been explained in terms of their relevance to both investigated motor tasks. The results precisely reflect the impulsive nature of motions studied through 942 A.Czaplicki et al. its consequences in the shape of curves of the torques and reaction forces. However, some caremust be taken when interpreting the results because they are sensitive to the procedure of smoothing the kinematic data. It also seems to be important to collect a large number of video frames, particularly during the supporting and landing phase of the motion. The obtained net torques may be used as the input data to solve an individual muscle distribution pro- blem by means of adequate optimization methods, which in turn allows one to compute ”real” reactions in the joints. The answer of which loads, and in which direction cross the joints during these activities may be of considerable interest to clinicians and trainers. Acknowledgments The support of this work by The Fundaçao para e Ciencia e Tecnologia, through the project EME/399976/Z, is gratefully acknowledged.The rawdata for the presen- ted study has been collected at The Faculty of HumanKinetics with the support by Dr. Orlando Jesus and Prof. Joao Abrantes, to whom the authors are grateful. References 1. Abdel-Aziz Y., Karara, H., 1971. Direct linear transformation from com- parator co-ordinates into object space co-ordinates,Proc. ASP/UI Symposium on Close-Range Photogrammetry, 1-18 2. Ambrósio J., Abrantes J., Lopes G., 2001, Spatial reconstruction of hu- man motion by means of a single camera and a biomechanical model, Human Movement Science, 20, 829-851 3. Blajer W., Czaplicki A., 2001, Modeling and inverse simulation of somer- saults on the trampoline, Journal of Biomechanics, 34, 12, 1619-1629 4. Blajer W., Czaplicki A., 2003, Contactmodeling and identification of pla- nar somersaults on the trampoline,Multibody System Dynamics, 10, 289-312 5. Bobbert M., van Schenau G., 1988, Coordination in vertical jumping, Jo- urnal of Biomechanics, 21, 3, 249-262 6. Celigüeta J.T., 1996,Multibody simulation of humanbodymotion in sports, Proc. XIV International Symposium on Biomechanics in Sports, 81-94 7. Eberhard P., Spägele T., Gollhofer A., 1999, Investigations for the dynamical analysis of humanmotion,Multibody System Dynamics, 3, 1-20 8. HaugE., 1989,Computer AidedKinematics andDynamics ofMechanical Sys- tems, Allyn and Bacon, Boston Biomechanical modelling for whole body motion... 943 9. HayJ.G., 1993,Citius, altius, longius (faster, higher, longer): thebiomechanics of jumping for distance, Journal of Biomechanics, 26, Suppl. 1, 7-21 10. Jalón, J., Bayo E., 1994, Kinematic and Dynamic Simulation of Multibody Systems: the Real-Time Challenge, Springer-Verlag, NewYork 11. Jalón J.G., Unda, J., Avello A., 1986, Natural coordinates for the com- puter analysis of multibody systems,Computer Methods in Applied Mechanics and Engineering, 56, 309-327 12. Laananen D., 1991, Computer Simulation of an Aircraft Seat and Occupant in a Crash Environment, Program SOM-LA/SOM-TA, DOT/FAA/CT-90/4, US Department of Transportation, Federal Aviation Administration 13. Laananen D., Bolukbasi A., Coltman J., 1983, Computer Simulation of an Aircraft Seat and Occupant in a Crash Environment - Volume 1, Techni- cal report DOT/FAA/CT-82/33-I, US Department of Transportation, Federal Aviation Administration 14. Lees A., Smith P.G., Fowler N.E., 1994, A biomechanical analysis of the last stride, touchdown and takeoff characteristics of the men’s long jump, Jo- urnal of Applied Biomechanics, 10, 61-78 15. 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 Bio- mechanics, 34, 1471-1482 16. Mieczkowski T., 1982,Akrobatics, WSP, Szczecin (in Polish) 17. Nikravesh P., 1988, Computer-Aided Analysis of Mechanical Systems, Prentice-Hall, Englewood Cliffs 18. PandyM.G., Anderson F.C., HullD.G., 1992,A parameter optimization approach for the optimal control of large-scalemusculoskeletal systems,Journal of Biomechanical Engineering, 114, 450-460 19. PandyM.,ZajacF., SimE.,LevineW., 1990,Anoptimal controlmodel for maximum-height human jumping, Journal of Biomechanics, 23, 12, 1185-1198 20. Schiehlen W., 1997, Multibody system dynamics: roots and perspectives, Multibody System Dynamics, 1, 149-188 21. Selbie W.S., Caldwell G.E., 1996, A simulation study of vertical jumping from different starting postures, Journal of Biomechanics, 29, 9, 1137-1146 22. Silva M., 2003, Human motion analysis using multibody dynamics and opti- mization tools, Ph.D. Thesis, Technical University, Lisbon 23. SilvaM.,Ambrósio J., 2002,Kinematic data consistency in the inversedyna- mic analysis of biomechanical systems,Multibody SystemDynamics, 8, 219-239 944 A.Czaplicki et al. 24. Silva M., Ambrósio J., Pereira M., 1997, Biomechanical model with joint resistance for impact simulation,Multibody System Dynamics, 1, 65-84 25. Winter D.A., 1990,Biomechanics and Motor Control of Human Movement, JohnWiley and Sons, NewYork 26. Zatsiorsky V.M., 1998, Kinematics of Human Motion, Human Kinetics, Champaign Biomechaniczne modelowanie trójwymiarowych ruchów człowieka w środowisku współrzędnych naturalnych Streszczenie W pracy zaprezentowano pełną metodologię do badania ruchów człowieka. Pod- stawowymelementemtejmetodologii jest trójwymiarowybiomechanicznymodel ciała ludzkiego.Położenia członówmodelu opisano za pomocąwspółrzędnychnaturalnych, cowyeliminowało konieczność użycia współrzędnych kątowych.Model składa się z 33 ciał sztywnychpołączonychprzegubami (stawami) i posiada 44 stopnie swobody.Wy- padkowemomenty sterującewposzczególnychstawachpochodząceod siłmięśniowych wprowadzonodo dynamicznych równań ruchumodelu za pomocą stosownychwięzów kinematycznych pomiędzy sąsiadującymi członami w danym stawie. Wartości tych momentówwyznaczonowykorzystując formalizmmnożników Lagrange’a.Model wy- korzystanodo identyfikacji wypadkowychmomentówmięśniowych i reakcjiwewnętrz- nych w głównych stawach kończyny dolnej człowieka podczas odbicia do skokuw dal oraz w trakcie wykonywania salta w tył z miejsca. Manuscript received March 19, 2004; accepted for print April 15, 2004