The main objective of this work is to overcome the most classical hypotheses used in kinematics (lower pair mechanical joints) and inverse dynamics computation (joints without resistance) including the estimation of muscular forces. Kinematics is addressed in the first part of the thesis by using “geometric” kinematic models consisting in simple elements (sphere, plane, shaft) modeling the anatomical structures. These models correspond to constraints in the kinematic computation (especially in multi-body optimization). The work consisted in introducing deformable ligaments by using a penalty-based method. It has beenshowed that this method used with a generic geometric model improved the estimation of the knee kinematics from the skin markers, when compared to more classical methods, and introduce physiological couplings between the degrees of freedom. Model personalization is also considered thanks to the flexibility of the method. The influence of the passive structure actions during gait is studied in the second part of the thesis. The work consisted in a local and a global study of those actions. The local study showed that the influence on the joint contact and musculo-tendon forces of the ligament passive moments is limited. The global study showed that the passive moments of the whole peri-articular structures contribute to the motor moments during gait and that the passive ligament moments available in the literature are not appropriate. The long term objective of those studies is to develop a multi-scale approach of the lower limb modeling. The proposed articular modeling (with more complex joints) allows a better interaction between the different scales of modeling (rigid multi-body vs. finite elements).
