Abstract
The ability to avoid obstacles is crucial for human mobility in most natural environments. Fajen and Warren (2003) proposed a steering dynamics model in which locomotor paths emerge on-line, without advanced planning. Goals act as attractors and obstacles as repellers of heading (the direction of travel), where heading is determined by the current eccentricities and distances of the goal and nearby obstacles ([[lt]] 4 m). Here we ask whether human routes are also affected by more distant obstructions (up to 8 m), implying more advanced planning. In addition, we investigate possible optimality principles for preferred routes. Participants walk in an immersive virtual environment to a goal post (13 m) while avoiding a near obstacle post (6 m) and a far barrier (8, 10 m or no barrier). Participants are instructed to take their preferred route to the goal, or one of the three possible routes through the obstacle course. If they anticipate the effect of a distant obstruction, their paths should be influenced by the presence and position of the far barrier. Measures on the observed paths assess whether the preferred path tends to minimize travel time, travel distance, and/or the total impulse (F × t) acting on the walker's center of mass. The results provide a test of the adequacy of the near-obstacle control model and suggest possible modifications to incorporate distant obstructions. In addition, they allow us to evaluate the properties of preferred locomotor paths that tend to be optimized, suggesting a basis for observed model parameters. A model that describes human paths through cluttered environments would prove useful in applications such as robot locomotion, pedestrian traffic engineering, and gait rehabilitation.
Reference: FajenBRWarrenWH (2003) Behavioral dynamics of steering, obstacle avoidance, and route selection. J. Exp. Psychol. Hum. Percept. Perform. 29:343–362.
Supported by NIH EY10923.