An Analytic Model To Evaluate The Influence of Uncertainty On The Cooperative Search Behaviors Of Autonomous UAVs Open Access Deposited
Downloadable ContentDownload PDF
We present an analytic framework for modeling and measuring uncertainty for the scenario of Unmanned Aerial Vehicles (UAVs) cooperatively searching for moving targets. Uncertainty exists in a UAV’s assessment of teammate locations, target locations, recently visited cells, and sensor results. As is frequently done, we use probabilistic maps to represent uncertain information regarding the UAV’s environment. We describe new methods to update the probabilistic maps when information arrives from onboard sensors or teammate UAVs. When new information is missing or delayed, we propose using a novel and straightforward diffusion approach to update probabilistic maps. The UAVs make navigation decisions based on response to potential fields generated by the probabilistic maps. Since map data have uncertainty, this leads to decisionmaking in uncertainty. We describe how uncertainty in the environment translates into a unique measure, velocity vector dispersion (DV), which describes the uncertainty in the UAV’s navigation decision. Thresholds related to DV may be useful to guide real-time decision policies. We present simulation results that show how the use of diffusion affects the time to locate targets. We also describe how DV varies during UAV flight and comment on its utility.