A reach motion generation algorithm based on posture memory

Abstract

Various models and algorithms (hereafter, simply algorithms) have been developed to simulate human motions. Most of these algorithms generate only a single “would-be-realistic” motion for a given scenario; thus, they cannot inform the designer of the full range of feasible human motions for the given scenario. In this paper, we present a novel reach motion generation algorithm based on the use of a posture memory, which aims to inform the range of feasible human reach motions for a given simulation scenario. In this algorithm, posture memory is constructed using a random posture generation and registration process. After memory construction, different paths connecting the starting and ending hand positions are created. Then, the human reach motion generation algorithm produces different “feasible” motions by selecting and connecting “connectable” postures found within the neighboring cells of the path. The algorithm proposed in this study generates feasible motions and an ability to generate and report the full range of feasible human motions for a given scenario allows a more complete understanding of the consequences of a design decision and also provides a basis for simulating human motions under different constraints.