Advancements and Challenges in Motion Control for Propulsive Autonomous Underwater Vehicles: A Systematic Review

Abstract

The propulsive autonomous underwater vehicle (AUV) serves as a critical technological asset for autonomous underwater missions, playing an increasingly vital role in ocean exploration, resource extraction, and environmental surveillance. However, achieving high-precision motion control remains a formidable challenge due to inherent system constraints, strong dynamic couplings, and the highly nonlinear, uncertain hydrodynamic environments encountered during operations. This paper provides a systematic survey of recent progress in motion control and optimization methodologies specifically tailored for propulsive AUVs. To establish a coherent foundation, we first delineate the canonical kinematic and dynamic models of representative AUVs and elaborate on effective motion decoupling strategies. Following the evolutionary trajectory of control paradigms, we then systematically categorize and discuss classical, nonlinear, intelligent, and optimization-based algorithms developed over the past two decades, with detailed emphasis on their applicative characteristics, modification logic, and experimental validation. Finally, the review identifies persisting research bottlenecks and articulates promising avenues for future investigation, grounded in practical implementation demands. By integrating current advancements and distilling critical insights, this review intends to offer a valuable reference and a clear roadmap for researchers navigating this complex domain. Note to Practitioners—This work is motivated by the practical difficulties encountered in controlling autonomous underwater vehicles (AUVs) during real-world ocean missions, where strong currents, sensor noise, system faults, and stringent energy constraints collectively undermine stable and precise operation. To advance motion control capabilities in such challenging operating conditions, we systematically review and compare commonly used control methodologies, ranging from classical PID regulators to advanced adaptive and learning-based frameworks, highlighting their practical strengths and limitations. For practitioners, the key insight is that no single method works best in all conditions. Instead, hybrid architectures offer a more reliable path to real-world deployment. Importantly, we identify major gaps: the preponderance of research remains confined to simulation-based validation, while critical real-world issues like coupled model mismatch and multi-physical constraints, stochastic ocean disturbances and coupled sensor uncertainty, embedded computational bottlenecks and real-time inferiority, and fault survivability and analytical redundancy for long-term autonomy are still underexplored. Future developments should prioritize integrated robust control with multi-objective collaborative optimization, computationally lightweight algorithms, multi-AUV collaboration, and built-in autonomous intelligence. These insights are broadly applicable to other autonomous systems operating in uncertain and resource-constrained environments, such as aerial drones and ground mobile robots.