PHYSIOLOGICALLY RELEVANT SIT-TO-STAND MOVEMENT WITH OUTPUT FEEDBACK TRACKING

Abstract

This research work proposes a modeling framework to simulate joints and head position trajectories as control variables during STS. Besides this, a modeling scheme for CNS’s inference mechanism to estimate appropriate joint angles and the torques to perform the required motion is also presented. Clinical studies suggest that CNS controls a motion and maintains the balance by gathering multisensory data. CNS develops short term motion control strategies, called fast dynamics, and long-term strategies called slow dynamics. Slow dynamics include learning the appropriate motion patterns. For any voluntary motion, CNS anticipates set patterns of inputs from multisensory systems, and compares them with patterns built in / learnt over long period of time, as slow dynamics. The concept of slow dynamics, is the motivation to develop a modeling and simulation framework, for the clinical hypothesis, that besides numerous other factors, CNS controls STS motion, by tracking pre-learned kinematic trajectories of joints and/or head position. The solution of this research problem is achieved through following steps: 1) STS motion data from young, able-bodied subjects are collected using infra-red cameras and passive reflective markers. A force platform is used to gather Ground Reaction Force (GRF) data during STS. 2) A biomechanical modeling and simulation framework to synthesize and control human-like STS using joint/head position trajectories is proposed, which comprises an analytical four-segment rigid body human biomechanical model, various controllers to model CNS as STS motion controller and the STS patterns learnt by CNS as reference trajectories/control variables. 3) Finally, the proposed modeling scheme is mapped on real subjects so that the synthesized motion may be compared with real humans’ STS motion. A comparison between experimental measurements and simulation results is used to validate motion synthesis frameworks. For low-level control linear robust controllers worked well. The task-level control was achieved by Adaptive Neuro-Fuzzy Inference System (ANFIS) based controller, although the ANFIS operation remains limited to feed forward control only. Cartesian control presents complete framework for task-level control but needs fine tuning for more realistic STS motion. The main achievements of this research work are 1) the development of STS motion and force dataset of healthy young adults. Thisviii dataset can be used as a reference to provide a comparison with some pathological STS. 2) The validation of a human model that was extensively used for analysis of STS over quite some time. 3) Customization of human biomechanical model on real subjects using weighing coefficient method 4) CNS modeling as robust controller, ANFIS and Cartesian controller using full/reduced order measurements 5) Task-space-training algorithm to train/customize ANFIS system. Accurate modeling and understanding of human motion have significant scope in the fields of rehabilitation, humanoid robotics and virtual characters' motion planning based on high-level task control schemes. In the future, we aim to study STS motion control based on brain signals using real subjects and compare our human-CNS modeling scheme used for the synthesis of the same motion.