Environment modeling for sense and avoid sensor safety assessment

Abstract

Sense and avoid (SAA) systems being developed for unmanned aircraft are needed to fulfill the requirement to see and avoid other aircraft. An environment model objectively describes a specific environmental condition so that the performance of unmanned aircraft SAA sensors can be accurately modeled in safety studies. This paper presents an approach to develop an environment model that encompasses elements of the environment that are external to the unmanned aircraft and influence sensor performance. The environment condition, as defined in this paper, consists of the atmosphere and the intruder aircraft signature. An environment model, used in conjunction with a sensor model, can be employed to demonstrate the expected overall performance of SAA sensors across millions of encounter situations. Bayesian networks constructed from a variety of data sources capture the statistical makeup of environmental conditions where an unmanned aircraft will operate. Using a Bayesian statistical technique ensures that important relationships between variables in the model are captured.