Specific requirements must guide the design of autonomous systems as they are increasingly present in our everyday environment. Their properties must be carefully defined and checked to guarantee safety, security and dependability. In this paper, we adopt a discrete event system modelling framework and focus on properties that are related to diagnosis. A new property called single state trackability is introduced. While available observations may lead to an ambiguous estimate , i.e. several admissible state candidates, this property assesses the possibility of reducing the estimate to a single state without this leading to a dead-end in the continuation of the execution. A single state estimate advantageously facilitates decision making and allows the use of a determin-istic planner in the autonomous architecture. We provide a necessary and sufficient condition for single state trackability of a discrete event system and we propose a recursive algorithm to check this property. The algorithm is validated with a set of benchmarks.