Evaluating Situation Awareness of Autonomous Systems

Abstract

Autonomous systems proved to be very successful in specialized problem domains. But their perception, reasoning, planning and behavior capabilities are generally designed to fit special purposes. For instance, a robotic agent perceives its environment in a way that was defined in advance by a human designer. The agent does not exhibit a certain perception behavior because it actually thinks it would be reasonable to do so. But with an increasing level of autonomy as well as a larger temporal and spatial scope of agent operation higher-level situation analysis and assessment become essential. This paper examines criteria for evaluating situation-awareness of autonomous systems and proposes methods to satisfy them. An example application scenario is presented that provides initial results for evaluating situation-aware systems.

TEAM FORMATION FOR AGENT COOPERATION IN LOGISTICS - Protocol Design and Complexity Analysis

“…A solution for such a case is to employ an interaction protocol with a lower degree of decentralisation (Schuldt, 2010). An alternative might be to restrict the number of participants, for instance, based on spatial regions of relevance (Gehrke, 2009).…”

Situation-Aware Environment Perception Using a Multi-Layer Attention Map

“…Albus [20], [21] introduced a layered architecture for task planning of autonomous agents in 1999, including modules for the inclusion of a-priori knowledge. Gehrke [22] introduced requirements a situation-aware system must meet that are derived from the levels of situationawareness from Endsley [23]. To the best of our knowledge, only one approach provides actual implementation details and performance results: Matzka et al [24], relating to Matzkas thesis from 2009 [25], introduced a situation-aware system for a forward-looking sensor setup that focuses on highway scenarios.…”