cs.ROMar 8, 2026

Underwater Embodied Intelligence for Autonomous Robots: A Constraint-Coupled Perspective on Planning, Control, and Deployment

Authors: Jingzehua Xu, Guanwen Xie, Jiwei Tang, Shuai Zhang, Xiaofan Li

Organizations: Department of Mechanical Engineering, The University of Hong Kong, Pokfulam Road, Hong Kong, China · Tsinghua Shenzhen International Graduate School, Tsinghua University, Shenzhen, 518055, China · Department of Data and Systems Engineering, The University of Hong Kong, Pokfulam Road, Hong Kong · Department of Data Science, New Jersey Institute of Technology, NJ 07102, USA

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Autonomous underwater robots are increasingly deployed for environmental monitoring, infrastructure inspection, subsea resource exploration, and long-horizon exploration. Yet, despite rapid advances in learning-based planning and control, reliable autonomy in real ocean environments remains fundamentally constrained by tightly coupled physical limits. Hydrodynamic uncertainty, partial observability, bandwidth-limited communication, and energy scarcity are not independent challenges; they interact within the closed perception-planning-control loop and often amplify one another over time. This Review develops a constraint-coupled perspective on underwater embodied intelligence, arguing that planning and control must be understood within tightly coupled sensing, communication, coordination, and resource constraints in real ocean environments. We synthesize recent progress in reinforcement learning, belief-aware planning, hybrid control, multi-robot coordination, and foundation-model integration through this embodied perspective. Across representative application domains, we show how environmental monitoring, inspection, exploration, and cooperative missions expose distinct stress profiles of cross-layer coupling. To unify these observations, we introduce a cross-layer failure taxonomy spanning epistemic, dynamic, and coordination breakdowns, and analyze how errors cascade across autonomy layers under uncertainty. Building on this structure, we outline research directions toward physics-grounded world models, certifiable learning-enabled control, communication-aware coordination, and deployment-aware system design. By internalizing constraint coupling rather than treating it as an external disturbance, underwater embodied intelligence may evolve from performance-driven adaptation toward resilient, scalable, and verifiable autonomy under real ocean conditions.

Figures & tables

Fig. 1: Application drivers and environmental constraints shaping underwater robotic autonomy . Representative application scenarios motivating the deployment of autonomous underwater robots include environmental monitoring, subsea resource exploration, and long-term ocean observation. These tasks require persistent operation, large-area coverage, and reduced human supervision. (a) Environmental monitoring of marine ecosystems using autonomous underwater vehicles [ 1 ] . (b) Subsea resource exploration and infrastructure inspection in complex underwater environments [ 7 ] . (c) Long-term ocean observation requiring sustained and wide-area data collection [ 10 ] . To support such missions, most existing systems adopt a modular autonomy architecture integrating perception, planning, and control for environment understanding and decision-making. However, reliable autonomy in real oceans remains challenged by several environmental constraints: (d) hydrodynamic uncertainty from time-varying currents and fluid–vehicle interactions; (e) perception degradation due to turbidity, light attenuation, and sensing limitations; and (f) sparse communication caused by bandwidth-limited and latency-prone acoustic channels.
Fig. 2: Structural distinctiveness of underwater embodiment compared to terrestrial and aerial systems. In terrestrial and aerial robotics, environmental interaction primarily acts as external disturbance [ 60 , 61 ] . By contrast, underwater systems operate within a dense, viscous, and dynamically coupled fluid medium that simultaneously reshapes controllability, observability, and communication reliability [ 62 ] . Hydrodynamic effects couple translational and rotational dynamics; optical and acoustic sensing quality depends on motion and environmental conditions; and acoustic communication introduces sparse, delayed coordination. These factors create recursive cross-layer coupling across perception, planning, and control while also constraining coordination through sparse and delayed communication. As a result, autonomy in underwater systems cannot be decomposed into independent modules without structural fragility.
Fig. 3: Embodied autonomy as constraint-coupled optimization in underwater systems. Rather than optimizing perception, planning, and control as isolated modules, underwater autonomy is more appropriately understood as a closed-loop, multi-objective regulation process over joint state, belief, and resource spaces. Mission utility, uncertainty regulation, and physical feasibility are structurally coupled: actions that improve task progress may increase localization drift, energy consumption, or dynamic instability, while uncertainty reduction and safety preservation can in turn reshape effective mission performance. The feasible policy manifold therefore represents the set of embodied strategies that balance these competing yet interdependent objectives, whereas sequential modular optimization is more likely to drive the system toward unstable or unsafe regimes under realistic ocean conditions.
Fig. 4: Application domains as structural stress tests of embodied autonomy. Representative underwater mission scenarios impose distinct yet partially overlapping constraint profiles that collectively challenge the stability and robustness of embodied autonomous systems. (a) Persistent monitoring places sustained pressure on long-horizon epistemic stability, requiring reliable state estimation and uncertainty management over extended deployments while simultaneously maintaining strict energy endurance constraints. (b) Infrastructure inspection emphasizes proximity-driven safety and dynamic feasibility, where operations near complex structures demand precise motion control, robust collision avoidance, and stable perception under turbulent hydrodynamic conditions. (c) Exploration magnifies the accumulation of epistemic uncertainty and the difficulty of maintaining globally consistent maps in previously unobserved environments, often under limited sensing and intermittent localization cues. (d) Multi-robot cooperation introduces additional system-level complexity through coordination sparsity, communication delays, and cross-agent belief fragmentation, requiring decentralized reasoning and bandwidth-aware information exchange. Although the dominant operational pressures differ across these domains, each scenario ultimately exposes the same underlying structural coupling across mission objectives, uncertainty regulation, physical feasibility, communication reliability, and long-term resource sustainability.
Fig. 5: Emerging research directions for resilient underwater embodied intelligence . Future underwater autonomy requires integrated strategies that regulate cross-layer coupling across perception, planning, and control, while explicitly accounting for communication, coordination, and deployment constraints. (a) Dreamer introduces a general model-based reinforcement learning algorithm that learns latent world models of environment dynamics and improves behaviour by imagining future trajectories, illustrating how physics-grounded world models can support anticipatory planning under uncertainty [ 179 ] . (b) OceanGPT presents a domain-specific large language model for ocean science that integrates heterogeneous oceanographic data and instruction-generation frameworks, highlighting the potential of foundation models grounded in ocean-domain knowledge [ 42 ] . (c) The Leash Actor–Critic (LAC) method proposes a constraint reinforcement learning framework that combines an emergency safety critic with Lagrangian optimization to restrict unsafe actions during AUV path planning, demonstrating how stability and safety constraints can be embedded within learning-enabled control [ 180 ] . (d) A collaborative USV–AUV system integrates Fisher-information–based localization with reinforcement-learning-based cooperative planning, illustrating scalable decentralized coordination under communication and environmental constraints [ 162 ] . (e) An open-source AUV path-planning benchmarking platform built on realistic underwater simulation environments provides standardized evaluation scenarios and metrics, highlighting the importance of field benchmarking and reproducible validation for underwater autonomy [ 181 ] .
More in this topic

Underwater Robotics

Underwater robotics develops autonomous and remotely operated machines for navigation, inspection, manipulation, and exploration in aquatic environments.

See all 148 papers

Explore similar work

CardsList
Sep 19, 2026

AquaCap: A Training-Free Underwater Embodied Agent with Code-as-Policy

Recent advances in vision-language-action models have stimulated growing interest in underwater embodied intelligence. However, their reliance on large-scale interaction data limits their applicability underwater, where data collection is costly and scarce. To address this challenge, we present AquaCap, a training-free Code-as-Policy framework for autonomous underwater navigation and manipulation. AquaCap employs a dual-layer agent that translates task instructions and environmental observations into condition-aware plans and executable control programs. Structured perception then provides the agent with semantic, geometric, and reliability-aware observations under degraded underwater conditions. A failure-aware memory diagnoses unsuccessful actions and supports closed-loop replanning and code revision. This design enables online adaptation without task-specific training or parameter updates. AquaCap achieves a 66.43% success rate in simulation. Real-world experiments further demonstrate autonomous grasping and object transport with an ROV, including the manipulation of targets displaced by hydrodynamic disturbances.

Xiaoshi Li, Yule Xu, Chunghiu Kong +5

Sun Yat-sen University · Dalian University of Technology · Southern Marine Science and Engineering Guangdong Laboratory

Underwater RoboticsAutonomous Underwater NavigationLanguage Model-Based Control
AskPDF
Aug 9, 2026

Hierarchical Topology-Aware Planning and Control of Underwater Vehicle-Manipulator Systems in Confined Environments

This paper addresses autonomous intervention with an underwater vehicle--manipulator system (UVMS) in confined, cluttered, and partially known environments, where poor maneuverability, narrow passages, and uncertain execution may cause the robot to enter unrecoverable regions. We propose MANTA, a three-layer hierarchical planning-and-control framework that couples passage accessibility, manipulation feasibility, and closed-loop execution. The first layer performs global connectivity reasoning in a conservative reduced base space to extract traversable corridor candidates toward the task region. The second layer refines each candidate corridor by jointly optimizing the continuous base motion and arm trajectory, producing a collision-free base--arm trajectory. The third layer learns a reach-and-hold base policy using Gaussian-process model-based reinforcement learning (MBRL) through MC-PILCO, enabling trajectory tracking and station keeping at the planned manipulation state. During execution, the framework monitors map updates and can trigger recovery and route repair when the active passage becomes infeasible. MANTA is evaluated in confined UVMS planning and closed-loop tracking experiments. Across 120 matched planning queries, it achieves higher task success than full-state sampling-based baselines while producing larger clearance margins and lower arm motion. The learned MC-PILCO policy further reduces position and yaw tracking errors on both training and unseen tube-like references. These results show MANTA as a structured and data-efficient framework for safe autonomous underwater intervention in caves, tubes, and cluttered subsea structures.

Mohamed Abdelwahab, Ruggero Carli, Damiano Varagnolo +1

Department of Information Engineering, University of Padova, Via Gradenigo 6/B, 35131 Padova, Italy · Department of Electrical and Information Engineering, Polytechnic University of Bari, Via Orabona 4, 70125 Bari, Italy · Department of Engineering Cybernetics, Norwegian University of Science and Technology, O. S. Bragstads plass 2D, 7034 Trondheim, Norway

Underwater RoboticsSafe Motion PlanningMotion PlanningAutonomous Underwater NavigationRobot Motion Planning
AskPDF
Apr 6, 2026

A Survey on Sensor-based Planning and Control for Unmanned Underwater Vehicles

This survey examines recent sensor-based planning and control methods for Unmanned Underwater Vehicles (UUVs). In complex, uncertain underwater environments, UUVs require advanced planning and control strategies for effective navigation. These vehicles face significant challenges including drifting and noisy sensor measurements, absence of Global Navigation Satellite System (GNSS) signals, and low-bandwidth, high-latency underwater acoustic communications. The focus is on reactive local planning layers that adapt to real-time sensor inputs such as SONAR and Inertial Measurement Units (IMU) to improve localization accuracy and autonomy in dynamic ocean conditions, enabling dynamic obstacle avoidance and on-the-fly re-planning. The survey categorizes the existing literature into decoupled and coupled architectures for sensor-based planning and control. The decoupled architecture sequentially addresses planning and control stages, whereas coupled architectures offer tighter feedback loops for more immediate responsiveness. A comparative analysis of coupled planning and control methods reveals that while PID controllers are simple, they lack predictive capability for complex maneuvers. Model Predictive Control (MPC) offers superior path optimization but can be computationally intensive, and invariant-set controllers provide strong safety guarantees at the potential cost of agility in confined environments. Key contributions include a taxonomy of architectures combining planning and control, a focus on adaptive local planning, and an analysis of controller roles in integrated planning frameworks for autonomous navigation of UUVs.

Shivam Vishwakarma, Tejal Bedmutha, Dharmendra Kumar Patel +2

Underwater RoboticsAutonomous Underwater NavigationRobotic Control
AskPDF