An Integrated Model of Gait and Transition Stepping for Simulation of Industrial Workcell Tasks University of Michigan

Format:

Conference/Event

Author/Creator:

Reed, Matthew P., author.

Conference Name:

2007 Digital Human Modeling Conference (2007-06-12 : Seattle, Washington, United States)

Language:

English

Physical Description:

1 online resource

Place of Publication:

Warrendale, PA SAE International 2007

Summary:

Industrial tasks performed by standing workers are among those most commonly simulated using digital human models. Workers often walk, turn, and take acyclic steps as they perform these tasks. Current human modeling tools lack the capability to simulate these whole body motions accurately. Most models simulate walking by replaying joint angle trajectories corresponding to a general gait pattern. Turning is simulated poorly if at all, and violations of kinematic constraints between the feet and ground are common. Moreover, current models do not accurately predict foot placement with respect to loads and other hand targets, diminishing the utility of the associated ergonomic analyses. A new approach to simulating stepping and walking in task-oriented activities is proposed. Foot placements and motions are predicted from operator and task characteristics using empirical models derived from laboratory data and validated using field data from an auto assembly plant. The motions of the pelvis and torso are predicted from the foot placements, operator characteristics, and task requirements. The lower-extremity motions are then generated using behavior-based inverse kinematics that relies on laboratory observations to address kinematic redundancy while respecting boundary constraints. This modular approach is highly general and can simulate gait, transition stepping, and stepping for balance maintenance in a single integrated system that can be implemented in any digital human model

Notes:

Vendor supplied data

Publisher Number:

2007-01-2478

Access Restriction:

Restricted for use by site license