Development of 3D Upper Body Posture Prediction Model for Static Lifting at Standing Posture

Abrishamkar, Amin | 2015

641 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 46890 (08)
  4. University: Sharif University of Technology
  5. Department: Mechanical Engineering
  6. Advisor(s): Arjmand, Navid; Zohoor, Hassan
  7. Abstract:
  8. One of the main causes of low back pain is mechanical loads acting on it. These loads can be measured by in vivo measurement using force and pressure sensors. Although this method leads to accurate results, it is invasive and costly. Currently the only noninvasive way to estimate these loads is to utilize biomechanical models. Using these models requires body posture as an input. Imaging techniques are suitable for determining the posture accurately, but these equipments are expensive and often impractical. So today posture prediction models are used as an alternative cost effective with acceptable accuracy. These mechanisms predict body posture using body mass, height, position and the amount of load in object's hand as inputs based on an optimization algorithm. In this study, human upper body has been modeled as a three-dimensional four bar linkage consist of forearm, upper arm and trunk in order to reach an optimized body posture for any given load magnitudes and positions. In addition to physiological and geometric constraints, an optimization algorithm is used to achieve optimal posture due to kinematic redundancy. The objective function used in this study is total joint torques and optimized code is written in MATLAB software. The results obtained from this model is compared and verified by postures predicted by 3DSSPP for 8 different activities. Qualitative and quantitative comparison of results indicates partial correspondence in some cases; however, the two postures do not match well in general because the type and the number of degrees of freedom defined for these two models and also the methods used to achieve the optimum posture are different. Moreover, we design a one-handed model while 3DSSPP is a full-body one. It should be noted that current model introduced in this project provides an inverse kinematic solution to estimate the posture while 3DSSPP postures are based on regression equations obtained from experimental data
  9. Keywords:
  10. Inverse Kinematics ; Optimization ; Genetic Algorithm ; Posture ; Lifting ; Posture Prediction ; Static Lifting

 Digital Object List