A GENERAL COMPUTER SIMULATION FOR THE ADAPTATION OF WALKING IN DIFFERENT OBSTRUCTED ENVIRONMENTS

Presented at NACOB 98:
North American Congress on Biomechanics
Canadian Society for Biomechanics - American Society of Biomechanics

University of Waterloo
Waterloo, Ontario, Canada
August 14-18, 1998

A GENERAL COMPUTER SIMULATION FOR THE ADAPTATION OF
WALKING IN DIFFERENT OBSTRUCTED ENVIRONMENTS

L. Galiana, B.J. McFadyen
Département de réadaptation, Université Laval
Québec, Canada, G1K 7P4

INTRODUCTION

To walk safely in our ever changing environment requires adaptive locomotor behaviour that successfully negociates environmental constraints. The purpose of this present work was to explore the ability of an existing simulation for low mid-swing obstacle avoidance to be generalized to situations involving higher and later obstacles as well as the accommodation to level changes.

REVIEW AND THEORY

During obstacle avoidance, the lower limb is re-organized to a knee flexor generation strategy to clear obstacles (McFadyen & Winter, 1991). However, during accommodation to higher floor levels there is an increase or maintenance of the hip strategy inherent to level walking (McFadyen & Carnahan, 1997). Such environmental adaptations have been termed anticipatory locomotor adjustments (ALAs).

A computer model developed by McFadyen et al. (1994) simulated the gait patterns for the avoidance of mid-swing obstacles of moderate height. Anticipated static states were related to the predicted limb adjustment required to clear the obstruction at its point of occurrence during the swing phase. Joint angles of the stereotypic level profiles were then modified with respect to this new static state through functions that weighted the percentage of adaptation at each lower limb joint. Zero percent maintained the level gait pattern at a given joint and 100% resulted in the full predicted static difference for that joint at a chosen point during the gait cycle. The timing and amplitude of adaptation for each joint could be set differently. Thus, the original anticipatory static state served only as a guide and not as a final outcome. This original model was shown to simulate lower limb dynamics for avoidance of small to moderately high, mid-swing obstacles.

PROCEDURES

In the present work, both a platform and an obstacle with heights of 20 cm were simulated as occurring in the latter part of the swing phase. Obstacle depth was set to 2 cm. The present results concentrated on ipsilateral control only. The inputs to the model were environmental conditions (position and height of the obstruction) and level walking kinematic data of one subject from McFadyen & Winter (1991). The future static guide was set as the point where the unmodified foot trajectory would first make contact with the obstruction. The weighting functions could be adjusted with respect to timing and amplitude returning to zero deviation at the end of the swing phase for obstacle avoidance or truncated to leave the limb in a flexed position when a new floor height was to be accommodated (see Figure 1). Iterations of model parameters were performed manually in order to find simulation results similar to previous experimental data.

Figure 1: Weighting Functions for obstacle (solid) and platform (dashed) conditions for the hip (thick) and knee (thin) joints.

RESULTS AND DISCUSSION

Figures 2 and 3 show the mechanical powers at the hip and knee for the obstacle avoidance and platform accommodation conditions respectively. The data resemble both kinematic and kinetic experimental data shown previously by McFadyen and Carnahan (1997) although amplitudes at the end of swing are somewhat higher. The simulation demonstrates that the concept of using a static guide of anticipated future states to adjust a set of joint weighting functions for adaptation of the joint angular kinematic patterns, can produce realistic looking results for both the avoidance of obstacles of different heights and the accommodation to higher levels. The generalization of such simple control supports the idea of an inverse kinematics organization at the planning and execution levels of locomotor control. This work has potential applications to robotic and active prosthetic control.

Figure 2 : Simulated trajectories (top), and hip (middle) and knee (bottom) power data during obstacle avoidance. Bars indicate extensor moments.

Figure 3 : As for Figure 2 for the accommodation to a new height.

REFERENCES

McFadyen, B.J. & Winter Neurosci. Res. Comm., 9, 37-44, 1991.

McFadyen, B.J., et al. Biol. Cybern., 72, 151-160, 1994.

McFadyen, B.J., Carnahan, H. Exp. Brain Res. 114: 500-506, 1997.

ACKNOWLEDGMENTS

We thank Mr. Guy St-Vincent for his technical help. This work was supported by a grant from NSERC.