Special Considerations for the Application of IVHM to Autonomous Aircraft and Vehicles
This SAE Aerospace Recommended Practice (ARP) provides guidance to develop and assure validation and verification of IVHM systems used in autonomous aircraft, vehicles and driver assistance functions. IVHM covers a vehicle, monitoring and data processing functions inherent within its sub-systems, and the tools and processes used to manage and restore the vehicle’s health.
The scope of this document is to address challenges and identify recommendations for the application of integrated vehicle health management (IVHM) specifically to intelligent systems performing tasks autonomously within the mobility sector. This document will focus on the core aspects of IVHM for autonomous vehicles that are common to both aerospace and automotive applications. It is anticipated that additional documents will be developed separately to cover aspects of this functionality that are unique to each application domain.
It is important to note that the scope of this document does not include systems that are allowed to continue independent training, learning and self-reconfiguration in the field.
Rationale: This Surface Vehicle & Aerospace Recommended Practice addresses the architecture and objectives for the use of Integrated Vehicle Health Management (IVHM) functionality in autonomous aircraft and vehicles. It will also identify the special considerations required for the succesful implementation, deployment and support of the IVHM functionality for these applications. This document may be applied by IVHM stakeholders to define the scope of the IVHM System in an autonomous application and how it should be integrated with other vehicle functions.
It is noted that this document does not fully address all aspects of the implementation, certification and support of the IVHM system or other vehicle systems but will provide a roadmap for use of related standards and recommended practices in which these details are provided.
For decades, engineers have designed systems to perform tasks autonomously (e.g., Autoland and TCAS), but the common aspect of these “traditional systems” is their behavior is well defined and bounded by the software/hardware configuration. The V&V process is straightforward due to the behavioral requirements restrained by the stimulus to the system and associated responses. Health management practices for these traditional autonomous systems have been well established and therefore not the focus of this document.
Intelligent systems require complex perception, reasoning, and decision making abilities not found in traditional systems. Their behavior is dictated by high level intent and having the ability to decompose intent into individual tasks. These expectations have driven the design of intelligent systems away from sequential software being executed on processors to novel architectures usually involving neural nets that require an additional training phase in the product development life cycle.
Traditional aero and AV applications that are envisioned will both have a deterministic response to the exact same inputs. Traditional aero applications have a much more deterministic environment allowing structured, procedural software. AV applciations will have a much more complex environment requiring non-traditional implementations.
It is apparent that the health monitoring of these novel architectures and their complex behavior has generated unique challenges that this document intends to identify and propose recommendations to maintain safe operation throughout their entire life cycle.