Validating an artificial intelligence human proximity operations system with test cases

Abstract

An artificial intelligence-controlled robot (AICR) operating in close proximity to humans poses risk to these humans. Validating the performance of an AICR is an ill posed problem, due to the complexity introduced by the erratic (noncomputer) actors. In order to prove the AICR’s usefulness, test cases must be generated to simulate the actions of these actors. This paper discusses AICR’s performance validation in the context of a common human activity, moving through a crowded corridor, using test cases created by an AI use case producer. This test is a two-dimensional simplification relevant to autonomous UAV navigation in the national airspace.