Design and Validation of a Synthetic Task Environment to Study Dynamic Unmanned Aerial Vehicle Re-Planning

Abstract

A key challenge facing unmanned aerial vehicle (UAV) operators is the need to re-plan routes on-the-fly when situations change. Operators must comprehend three-dimensional (3D) scenes and a multitude of potentially competing 3D mission and routing constraints in order to successfully re-plan, often under time pressure. Currently, there is significant research interest in supporting UAV operators through automation and improved visualizations. However, development and integration of these methods requires a careful understanding of the 3D spatial awareness challenges and requirements facing operators. To facilitate this understanding, here we report the design and validation of a synthetic task environment (STE) and testbed to study UAV re-planning. The STE is derived from a recent task analysis conducted with Navy UAV operators that focused on the key 3D spatial challenges entailed in re-planning. In an initial validation of the STE implemented in a re-planning testbed, several measures of re-planning performance were assessed for 36 participants working through controlled re-planning scenarios. The presence of mountainous terrain and the spatial overlap of mission constraints were parametrically varied. Performance was consistently worse in mountainous terrain, and in more highly-constrained conditions in mountainous terrain. In flat terrain, however, less constrained conditions resulted in paradoxically worse performance. Results have both basic and applied implications. Theoretically, the study provides a bridge between applied re-planning research and classic human problem solving work by allowing apparently simpler, unconstrained re-planning to be conceived of as less bounded search through re-planning problem space. For application, the results help constrain and define the requirements for future 3D visualization and automation support for UAV re-planning displays.