Investigating Locomotion Performance Using Evolutionary Robotics (ER) Based Simulation Program and Finite Element Analysis (FEA)

Abstract

Locomotion is a functional measure of the walking ability of vertebrates. It involves parameters like maximum running speed and gait analysis. Gait analysis has been previously employed using parameters like bone scaling, strength, calculating safety factor and typical ground reaction forces (GRFs). Other approaches rely on developing musculoskeletal models of vertebrates in reference to the locomotion patterns. Recently, scientists have developed forward dynamic evolutionary robotics simulation program (GaitSym) led by a research team at the University of Manchester UK. This program is used to study locomotion and gait using input parameters like muscle mass and joint coordinates with the lowest possible metabolic cost. The major advantage of this program is that we can understand how vertebrates moved in the past utilizing their size to function features based on animal’s morphology. However, the disadvantage is that we do not still understand how large or small magnitude of force is required to execute a reasonable gait pattern. In this research, we have generated computer models of gait analysis using a vertebrate model (Phasianus colchicum). The uniqueness of this study lies in a fact that the analysis was performed outside evolutionary robotics simulation program (i.e. GaitSym) using the same strategy as in the GaitSym. However additionally, Finite Element Analysis was also incorporated to overcome the GaitSym’s inability to consider bone’s geometry. This has provided us an opportunity to integrate parameters which was otherwise not possible in GaitSym so as to acquire realistic gait patterns. This study also utilized laws of mechanics for proper definition of gait boundary condition by utilizing correct vectors. Using image based-finite element analysis (FEA) boundary conditions were applied for both phases of gait i.e. stance phase and swing phase. Results from both phases were compared to investigate realistic magnitude of force parameters to address the overall vertebrate‘s skeleton integrity which was found reasonable. FE results acquired from this study are promising compared to the results available in literature to-date. Stresses acquired from stance phase and swing phase are 16MPa and 99MPa respectively. These stresses are a bit under estimated yet within the safety range of bone. Whereas the ultimate compressive strength of bone is reported as 15 180MPa – 200MPa (Currey, 2002). However both models (Stance and Swing phase) depict a reasonable metabolic cost. Moreover the deformation models with varying mesh densities have also been reasonably converged. It is concluded that GaitSym may hold true for existing exoskeletons to a certain limit knowing no geometry is considered. However there is a need to optimize the genetic algorithm of GaitSym for balanced locomotion. This is only possible if GaitSym alters its input force in the ODE or optimize its coordinates of skeleton relative to the simulations conducted in this study to retain skeleton integrity. A novel synergy for incorporation of stress into GaitSym has also been proposed at the end of this research.