Synthetic vision systems: human performance assessment of the influence of terrain density and texture

K Lemos; T Schnell; T Etherington; T Vogl; A Postikov

doi:10.1109/DASC.2003.1245924

The implementation of Synthetic Vision Systems (SVS) has posed a number of design questions. One of these questions is centered on the minimum required resolution of the Digital Elevation Model (DEM) database. Although a very detailed database may be desirable from a product appeal point of view or for improved representation of the elevation information., there simply may not be enough processing power available to render the high number of polygons with a texture at a high frame rate. Also, there may be a marginal rate of return of increasing DEM resolution when it comes to pilot performance, workload and situational awareness (SA). Another important issue is the choice of texture used on the synthetic terrain images. In the past, researchers and designers have debated about the relative merits of different textures (e.g., photo-realistic, elevation coloring, checkerboard). A third issue concerns the shading model, which may influence performance or SA through the potential for depth perception cues. The Operator Performance Laboratory (OPL) at the University of Iowa conducted a series of three experiments to assess human performance as a function of DEM resolution, terrain texture and shading methods. The DEM resolutions that were studied ranged from 3 arc seconds (best resolution) to 30 arc seconds (worst resolution). Textures included wire-frame (green on black), checkerboard, elevation coloring, contour lines, photo-realistic, and a non-textured plain brown color. Shading models included no shading, Gouraud shading, and flat shading. Part I involved an image identification task, in which the ability of non-pilot participants to recognize terrain features for static (Experiment 1) and dynamic terrain images (Experiment 2) was measured. Part II (Experiment 3) measured pilot performance by cross track error when na The implementation of Synthetic Vision Systems (SVS) has posed a number of design questions. One of these questions is centered on the minimum required resolution of the Digital Elevation Model (DEM) database. Although a very detailed database may be desirable from a product appeal point of view or for improved representation of the elevation information., there simply may not be enough processing power available to render the high number of polygons with a texture at a high frame rate. Also, there may be a marginal rate of return of increasing DEM resolution when it comes to pilot performance, workload and situational awareness (SA). Another important issue is the choice of texture used on the synthetic terrain images. In the past, researchers and designers have debated about the relative merits of different textures (e.g., photo-realistic, elevation coloring, checkerboard). A third issue concerns the shading model, which may influence performance or SA through the potential for depth perception cues. The Operator Performance Laboratory (OPL) at the University of Iowa conducted a series of three experiments to assess human performance as a function of DEM resolution, terrain texture and shading methods. The DEM resolutions that were studied ranged from 3 arc seconds (best resolution) to 30 arc seconds (worst resolution). Textures included wire-frame (green on black), checkerboard, elevation coloring, contour lines, photo-realistic, and a non-textured plain brown color. Shading models included no shading, Gouraud shading, and flat shading. Part I involved an image identification task, in which the ability of non-pilot participants to recognize terrain features for static (Experiment 1) and dynamic terrain images (Experiment 2) was measured. Part II (Experiment 3) measured pilot performance by cross track error when navigating with sole reference to SVS terrain images, following a nominal path along the terrain model. Both of these performance measures suggest a relative level of terrain awareness. Several measures of workload were also recorded: response time (Experiment 1), and ease of navigation (Experiment 3). The combined results from this series of investigations suggest that performance is highest and workload is lowest with a DEM resolution of 3 and 6 - arc second images, with Gouraud shading and with the checkerboard or elevation-coloring textures. Photo-realistic images were associated with a higher workload for image identification tasks, and the additional visual detail in the photo-realistic images did not increase performance for either identification or pilot performance tasks. Finally, higher workload levels for pilots were associated with wire-frame textured images. This paper presents these findings, with a focus on results for pilot performance and workload.

Synthetic vision systems: human performance assessment of the influence of terrain density and texture

View Online

Abstract

Details

Metrics