On the manipulability of swing foot and stability of human locomotion

Abstract

Manipulability is a measure that quantifies the range of possible motions of a robotic end-effector and it is also an important measure in the study of coordination of human upper body and grasping tasks. This measure, which is defined on both the kinematic and dynamic level, could be useful in gait, as it could be used to determine potential foot placement possibilities. Kinematic manipulability is defined based on the Jacobian and dynamic manipulability on both the Jacobian and mass-inertia matrix. The main purpose of this study was to evaluate the manipulability of human walking and to explore a possible relation between the manipulability and dynamic stability of walking at different speeds. A 37-DoF tree-like model of the human body was developed to evaluate the manipulability index of human walking. We measured kinematics of 11 healthy male subjects while walking on a treadmill, and mapped the data to the model using inverse kinematics. Jacobian based kinematic/dynamic manipulability measures of walking were evaluated for the swing phase of walking. Manipulability ellipsoids were drawn for geometric determination of this measure in all directions during early, mid-and late swing phases. As stability metrics, the local divergence exponent and Floquet Multipliers were calculated. The results indicated a high kinematic manipulability of the swing foot during early and late swing phases and a drop in kinematic manipulability during mid-swing. Kinematic manipulability of the swing leg during early and late (but not mid-) swing phases increased with walking speed but the average kinematic manipulability of the center of mass and dynamic manipulability of swing foot decreased with increasing walking speed. Moreover, the results showed a weak relation between the manipulability and local and orbital stability.

Online Adaptive PID Control for a Multi-Joint Lower Extremity Exoskeleton System Using Improved Particle Swarm Optimization

“…In previous research, the movement of the exoskeleton is always divided into the swinging phase and the stance phase, and they are modeled separately [18,[45][46][47]; therefore, it is not necessary to directly describe the foot-ground interaction and the constraint force of the knee joint after the leg is straightened. In this research, the dynamics of the exoskeleton in a complete gait cycle are considered, and two strong nonlinear factors, i.e., the foot-ground interaction and the constraint force at the knee joints, are modeled and incorporated.…”

“…However, generally, the system is too complicated to be modeled accurately, and previous research mainly focused on developing equivalent models by more or less simplifying or neglecting certain factors for controlling purposes. For example, the human body, which is fundamentally a complex musculoskeletal system, is always modeled into a linked multi-rigid-body system [14][15][16][17][18]; the human-robot interaction (HRI), due to its persistent and unpredictable nature, is usually simplified in the model [19,20]; the dynamics of the exoskeleton can also be described by a first-order transfer function [19,21] or a data-driven model [22], while the complexity will exacerbate as the DOF increases.…”

“…where the proportional and integral gains remain the same as Equation (18). In what follows, unless otherwise stated, the search space defined in Equations ( 17) and ( 18) also apply to the other cases.…”