System-Level Autonomy Trust Enabler (SLATE)

System-Level Autonomy Trust Enabler (SLATE)
Abstract

The presence and use of Unmanned Autonomous Systems (UASs) is increasing. There are more UASs, with greater levels of autonomy, executing missions of increasing complexity and criticality, resulting in an increased need for testing and verification of UAS system behavior. Unfortunately, the complexity of UAS hardware, software, environment, and mission profiles make testing coverage for system-level behavioral properties prohibitively difficult. This paper describes the System-Level Autonomy Trust Enabler (SLATE). SLATE employs constraint-based models to reason about the relationship between component-level behavioral guarantees and system-level properties for highly-autonomous systems interacting with a complex environment. Originally targeted for application to autonomous spacecraft and planetary rovers, SLATE provides unique capabilities for composing component-level guarantees into system-level guarantees, including inference across the abstractions dividing different layers in the control system architecture. Conversely, given a desired system-level property, SLATE can be used to derive a sufficient set of component level (and thus testable) properties such that the system property can be shown to hold. This paper describes the SLATE prototype, the process and set of tools used, our system models and performance data showing SLATE in operation, and the advantages of system-level T&E using composition of component properties both within and across abstraction layers.

Year of Publication
2008
Source
2008 U.S. Air Force T&E Days