The use of visual occlusion as a cue to altitude maintenance in low-altitude flight (LAF) was investigated. The extent to which the ground surface is occluded by 3-D objects varies with altitude and depends on the height, radius, and density of the objects. Participants attempted to maintain a constant altitude during simulated flight over an undulating terrain with trees of various heights, radii, and densities. As would be predicted if participants used occlusion, root-mean-square error was related to the product of tree height and tree density (Experiment 1) and to the product of tree radius and tree density (Experiment 2). This relationship was also found for simulated terrains with a more realistic mixture of tree heights (Experiment 4). The authors present a modification to an occlusion model (T. Leung & J. Malik, 1997) that can be used to approximate occlusion in the context of LAF, and they evaluate the modified model using the present LAF data. On a practical level, simulating 3-D objects is computationally expensive. The present results suggest that performance may be maintained with fewer objects if their size is increased.
Keywords: aviation, perception and action, visual occlusion, simulationThe ability to effectively control movement through the environment is one of the most important perceptual-motor skills humans possess. Whether walking, cycling, driving, or flying we must control our rate and direction of travel in a manner that will allow us to reach our goal while avoiding collisions with obstacles. Safe performance of these actions frequently requires that we regulate our position relative to some object or feature in the environment-for example, maintaining a safe distance behind a lead car in driving or keeping a safe distance from the road edge when cycling. In the present article we consider a particular example of this ability: regulating height above the ground in simulated aircraft flight.Of all of the perceptual-motor tasks performed by military pilots, low-altitude flight (LAF) is one of the most demanding and potentially the most dangerous. LAF can involve maintaining an altitude of less than 40 m while traveling at speeds up to 232 m/s (450 knots). Not surprisingly, this flight task accounts for disproportionately high numbers of accidents relative to the total number of flight hours (Wiener, 1988). Given this high level of risk, flight simulators are now used extensively to provide a training environment for pilots to acquire LAF skills.In order for simulator training of any flight skill to transfer positively to the real world, it is critical that the simulation include the necessary visual cues. Owing to both computational and display limitations, current flight simulators cannot provide all of the cues associated with LAF. The questions then become, to what extent can LAF tasks be performed given the cues that are available, and to what extent can those cues be traded off so as not to exceed limited simulator capabilities? Answering these questions requires empirical research...