Miguel Pedro Silva

Kinematic Data Consistency in the Inverse Dynamic Analysis of Biomechanical Systems

Inverse dynamic analysis is used in the study ofhuman gait to evaluate the reaction forces transmittedbetween adjacent anatomical segments and to calculate thenet moments-of-force that result from the muscle activityabout each biomechanical joint. The quality of theresults, in terms of reaction and muscle forces, is greatlyaffected not only by the choice of biomechanical model butalso by the kinematic data provided as input. This three-dimensional data is obtained through the reconstruction ofthe measured human motion. A biomechanical model isdeveloped representing human body components with acollection of rigid bodies interconnected by kinematicjoints. The data processing, leading to the spatialreconstruction of the anatomical point coordinates, usesfiltering techniques to eliminate the high frequencycomponents arising from the digitization process. Thetrajectory curves, describing the positions of theanatomical points are obtained using a form of polynomialinterpolation, generally cubic splines. The velocities andaccelerations are then the polynomial derivatives. Thisprocedure alone does not ensure that the kinematic data isconsistent with the biomechanical model adopted, becausethe underlying kinematic constraint equations are notnecessarily satisfied. In the present work, thereconstructed spatial positions of the anatomical pointsare corrected by ensuring that the kinematic constraints ofthe biomechanical model are not violated. The velocity andacceleration equations of the biomechanical model are thencalculated as the first and second time derivatives of theconstraint equations. The solution to these equationsprovides the model with kinematically consistent velocitiesand accelerations. The procedures are demonstrated throughthe application to a normal cadence stride period and theresults discussed with respect to the underlying principlesof the techniques used.

Sensitivity of the results produced by the inverse dynamic analysis of a human stride to perturbed input data

The results of the inverse dynamic procedures used in gait analysis are known to be highly dependent on the quality of the kinematic and dynamic input data and on the biomechanical model anatomical data. In this paper the sensitivities of the system response to imprecision in the input data and biomechanical model were calculated. It was shown that the gait analysis results were very sensitive to the identification of the point of application of the external forces. The quality of the results was less sensitive to errors made during motion reconstruction and to uncertainties in the biomechanical anatomical data. In this study it is also shown that the adopted inverse dynamic analysis method, based on natural coordinates, effectively shielded any error made on a particular kinematic chain from propagation to other branches of the biomechanical model.

Biomechanical Model with Joint Resistance for Impact Simulation

Based on a general methodology using naturalco-ordinates, a three-dimensional whole body responsemodel for the articulated human body is presented inthis paper. The joints between biomechanical segmentsare defined by forcing adjacent bodies to share commonpoints and vectors that are used in their definition.A realistic relative range of motion for the bodysegments is obtained introducing a set of penaltyforces in the model rather than setting up newunilateral constraints between the system components.These forces, representing the reaction momentsbetween segments of the human body model when thebiomechanical joints reach the limit of their range ofmotion, prevent the biomechanical model from achievingphysically unacceptable positions. Improved efficiencyin the integration process of the equations of motionis obtained using the augmented Lagrange formulation.The biomechanical model is finally applied indifferent situations of passive human motion such asthat observed in vehicle occupants during a crash orin an athlete during impact.

Solution of Redundant Muscle Forces in Human Locomotion with Multibody Dynamics and Optimization Tools

Abstract The analysis of human motion using inverse dynamics approaches allows for the evaluation of the muscle forces based on the knowledge of the kinematics of the human subject and the applied external forces. Owing to the redundant nature of the muscle arrangement, the system of equations available for the solution of the inverse dynamics problem has more unknown muscle forces than available equations. Therefore static optimization techniques are required to obtain a solution of the problem. Emphasizing applications to gait analysis and normal sports activities a whole-body, three-dimensional, biomechanical model of the human muscle-skeletal system is purposed to support the inverse dynamic analysis. In its current state of development, the biomechanical model includes a detailed description of the principal muscles of the locomotion apparatus for the lower limbs while net moments-of-force are used to represent the lumped muscle action about each other anatomical joint of the model. The muscles used in the biomechanical model include point-to-point and wrap-around muscles, depending in their nature and function, which results in a geometrically realistic muscle arrangement. A solution procedure for the redundant problem based on the static optimization is proposed here. Different types of objective functions representing the muscle forces, joint reaction forces, energy, or combinations of these are used in alternative. The muscle contraction dynamics is included in the optimization problem through the application of the Hills muscle model. The methodology developed is applied to a case of gait analysis with normal cadence. The results obtained are discussed in face of the modeling assumptions used in the biomechanical model and muscle system representation.

A Biomechanical Multibody Model with a Detailed Locomotion Muscle Apparatus

The ability of the animals to repeat the same movement or posture by recruiting different muscles or by using different muscle activation patterns is well known. From the mathematical point of view the solution of this problem and consequent determination of the recruited set of muscles and associated forces involves the solution of an optimization problem, in which the intrinsic objectives used by the central nervous system to recruit the referred set of muscles are represented by means of proper physiological cost functions. The objective of this work is to present a multibody dynamics based methodology to model the human body and the relevant features of the locomotion apparatus required for gait. For this purpose, a whole-body biomechanical model is used within the framework of an inverse dynamic analysis formulation with fully Cartesian coordinates, to calculate the individual muscle forces in the locomotion apparatus, the net moments of force at the joints of the upper body and the joint reaction forces developed between the anatomical segments of the biomechanical model when performing the specified task. Myoactuators representing the most relevant muscles of the locomotion apparatus are introduced using a Hill-type muscle model. Different cost functions are used to represent the objectives of the central nervous system when developing a particular task for the gait cycle. Sequential quadratic optimization tools are used to resolve the force-sharing problem arising from having a number of unknowns, associated to the individual muscle forces, higher than the number of available equations of motion, representing the dynamics of the anatomical segments that represent the human body. The methodologies proposed here are applied to a normal cadence gait cycle. In the process the most suitable cost functions for the specific task under analysis are identified and the quality of the results produced for this type of indeterminate problems is discussed.

Solving real-time scheduling problems with Hopfield-type neural networks

Real-time applications are increasingly becoming more complex, leading to the necessary development of fast scheduling algorithms. Therefore, the use of algorithms with a parallel search of feasible schedules seems to be attractive. In turn, Hopfield-type neural networks are suitable to solve complex combinatorial problems, owing to their fast convergence, if analog hardware is implemented. However, these neural networks have associated concepts of sub-optimality and the possibility of unfeasible solutions, which are contrary to the notion of system predictability. The paper presents a systematic procedure to map the scheduling problem onto a neural network in such a way that network solutions are always feasible schedules. Network convergence time is studied with digital computer simulations, using a discrete time model. Global asymptotic consistency between the discrete time model and the continuous one is assured. The paper also presents an analysis of the complexity of the proposed method.

Mosaic based flexible navigation for AGVs

Highly flexible systems require that Automatic Guided Vehicles (AGVs) in a plant navigate autonomously and changes in their missions should not require difficult setup procedures. In this paper we address the problem of localization of a mobile robot in a indoor environment. The robot is able to find its position without any grounded wires, landmarks or laser beacons. The robot uses images acquired in the roof to compute its position and navigate between coordinates. The main contribution is the absence of external services to solve the AGV localization problem, allowing fast reconfiguration.

Decentralized formation control of autonomous mobile robots

Control of vehicle formations is an area of great interest that requires knowhow from several scientific areas such as control theory, vision, communication system, etc. In this work, several approaches to maintain a formation of vehicles are explained and tested. In this paper, the approach towards formation control is achieved through Position Based Visual Servoing (PBVS). In PBVS, a cartesian space controller is used for target tracking. Two 2D pose estimation methods are described and two local controllers developed (Discrete Linear Quadratic Regulator and an explicit Model Predictive Controller). To maintain the mobile robots in formation, a Neighbor Referenced (NR) control scheme is used. NR-based control allows decentralized formation control schemes since each member of the formation only has information of its neighbors and not of the entire robots in the formation. Finally, simulation and experimental results of the developed work are presented.

Autonomous mobile robots localization with multiples iGPS Web Services

Diverse systems have been proposed to aid the navigation of mobile robots in indoor environments. Simultaneously, communication technologies that allow the integration of diverse systems even when these are developed on different platforms, have been proposed.