Background: Most existing models/algorithms for simulating goal-directed human motions were designed to generate a single "realistic" motion for a given input scenario.
Objective: This study presents a novel reach motion generation algorithm utilizing multiple posture memories. The algorithm aims to compute and visualize a set of human reach motions that approximates the full range of physically and physiologically feasible human motions for a given input scenario.
Methods: The algorithm utilizes posture memories constructed specifically for an individual worker using a probabilistic posture generation and registration process. The posture memories relate a hand position to the set of postures that place the individual's hand in its vicinity. When given an input scenario, the algorithm first generates different hand paths connecting the starting and ending hand positions specified in the scenario. Then, for each hand path, the algorithm produces different "feasible" motions by selecting and connecting multiple postures stored in the posture memories; the postures corresponding to the hand positions along the hand path are utilized.
Conclusions: The proposed algorithm helps understand the impacts of workplace design on the range of feasible human motion behaviors, and, thereby, contributes to the computer-aided ergonomics design of work tasks and workplaces.
Keywords: Digital human modeling; computer-aided ergonomics; human motion simulation; reachability; virtual ergonomics.