A Comprehensive Review of Parallel Autonomy Systems Within Vehicles: Applications, Architectures, Safety Considerations, and Standards

With traditional autonomy systems still far from achieving level 5 autonomy, Parallel Autonomy (PA) systems demonstrate increased potential to enable driving for all, regardless of the driver’s ability to perform the dynamic driving task. Their major goal is not to take away driving responsibilities from the driver but to augment the driver’s skills as needed to build safe and reliable autonomy systems. This paper presents a comprehensive review of PA systems, exploring their key concepts, architectures, functionalities, and existing challenges. Terminology related to PA system and how it is different from the concepts like Mixed Autonomy and Blended Control have been addressed first. A review of relevant standards follows, highlighting the current gap in specific PA regulations and identifying potential references from existing standards in related domains. The core of the paper delves into the architecture and concepts of PA systems. The paper discusses the various levels of autonomy applicable to PA, explores the different subsystems involved (perception, planning, control, human-machine interface), and analyzes hardware and software architecture considerations. There is also an examination of the research exploring decentralized control approaches for robust and safe PA systems. Safety considerations and Verification & Validation (V&V) methods are crucial aspects addressed in this review. It presents existing safety standards and considers customized test benches and V&V approaches from research projects. While specific findings and proof-of-concept implementations may be included depending on available information, the main focus lies on presenting a comprehensive overview of the current state of PA systems and identifying key areas for future research and development, especially the topics that would need guidelines and standardization related to PA systems.