In this paper, we propose a novel cross-platform fault-tolerant surfacing controller for underwater robots, based on reinforcement learning (RL). Unlike conventional approaches, which require explicit identification of malfunctioning actuators, our method allows the robot to surface using only the remaining operational actuators without needing to pinpoint the failures. The proposed controller learns a robust policy capable of handling diverse failure scenarios across different actuator configurations. Moreover, we introduce a transfer learning mechanism that shares a part of the control policy across various underwater robots with different actuators, thus improving learning efficiency and generalization across platforms. To validate our approach, we conduct simulations on three different types of underwater robots: a hoveringtype AUV, a torpedo shaped AUV, and a turtle-shaped robot (U-CAT). Additionally, real-world experiments are performed, successfully transferring the learned policy from simulation to a physical U-CAT in a controlled environment. Our RLbased controller demonstrates superior performance in terms of stability and success rate compared to a baseline controller, achieving an 85.7 percent success rate in real-world tests compared to$\mathbf{5 7. 1}$percent with a baseline controller. This research provides a scalable and efficient solution for fault-tolerant control for diverse underwater platforms, with potential applications in real-world aquatic missions.

This work explores the use of artificial whiskers as tactile sensors for enhancing the perception and navigation capabilities of mobile robots in challenging settings such as caves and underground mines. These environments exhibit inconsistent lighting conditions, locally self-similar textures, and general poor visibility conditions, that can cause the performance of state-of-the-art vision-based methods to decline. In order to evaluate the efficacy of tactile sensing in this context, three algorithms were developed and tested with simulated and physical experiments: a wall-follower, a navigation algorithm based on Theta*, and a hybrid approach that combines the two. The obtained results highlight the efficacy of tactile sensing for wall-following in intricate environments. When paired with an external method for pose estimation, it further aids in navigating unknown environments. Moreover, by integrating navigation with wall-following, the third, hybrid algorithm enhanced the map traversal speed by roughly 26−43% compared to standard navigation methods without wall-following.

Compliant fish-like robots are being developed as efficient and dependable underwater observation platforms with low impact on the observed environment. Orientation control is an essential building block to achieve autonomy for those vehicles. So far, the major focus has been on rigid tails or on flexible tails with a high degree of actuation. We present a novel control strategy for an underactuated robotic fish with a flexible tail optimized for cruising. The basis for our approach is the generation of asymmetric velocity profiles of the robot's tail beats. To achieve such velocity profiles, the usual sinusoidal tail actuation is replaced with skewed triangle waves. We provide a simple formulation for such waves, where their skew is dependent on only one variable which we define as skew factor. Furthermore, a nonlinear control law is derived to achieve the desired turning motions. We implement the controller on a compliant fish-like robot with a simple actuation mechanism. The control scheme is experimentally validated, and its robustness is tested in field trials.

Classification of river morphology is often based on hydromorphological units (HMUs) identified from field measurements. Established survey methods rely on expert judgment or collection of field point measurements. When used for HMU classification, these methods can suffer from high errors due to the variations in the sampling environment, causing low repeatability. In order to expedite field data collection and increase HMU classification accuracy, we propose a multisensory device, the hydromast. Each hydromast provides a new source of data to classify HMUs. The modules are inexpensive and highly portable, consisting of a synchronous array of commodity pressure and inertial sensors. Rapid, local changes in the flow field are recorded with absolute and differential pressure sensors. At the same time, slower depth-integrated flow signals are obtained from a small damped cylindrical mast, driven by vortex-induced vibrations. In contrast to existing passive flow measurement technologies, the hydromast uses fluid-body interactions to provide flow measurements. This allows for minimal signal processing and simple feature extraction. An array of three hydromasts was used to collect ten samples in three river HMUs with shallow depths and highly turbulent flows with smooth and rough beds. We investigated classification accuracy using single, dual, and triple hydromast arrays with pressure, inertial, and combined features using linear regression, a genetic algorithm, and a neural network. Although limited in scope, the set of spot measurements covering three HMUs showed that a single multimodal sensor could deliver an overall classification accuracy of 89% of the HMUs, and an increase of up to 99% was achieved using a multimodal triple hydromast array. These preliminary results show promise in using hydromasts for rapid and robust HMU classification, providing a new way to collect and assess river survey data.

Control of underwater vehicles is a thoroughly investigated subject but still an open problem, because of the environmental disturbances, the highly nonlinear behaviour of vehicles, the complexity of the vehicle hydrodynamics, etc. In this paper, we are interested in depth control of a bioinspired U-CAT underwater AUV in real operating conditions. Two depth control schemes are proposed, including a PID controller and a nonlinear RISE feedback controller. The proposed controllers are implemented on the robot, then tested in an open water environment. The obtained results are presented and discussed through different experimental scenarios to illustrate the efficiency of the proposed controllers, not only to successfully control the depth, but also to be robust towards external disturbances and parameters uncertainties. we conclude that RISE controller is more robust towards environmental disturbances and outperforms the PID controller when the robot is tested in real operating condition.

This paper demonstrates a novel motion control approach for biomimetic underwater vehicles with pitching fins. Even though these vehicles are highly maneuverable, the actuation of their different degrees of freedom (DOFs) is strongly coupled. To address this problem, we propose to use smooth DOF prioritization depending on which maneuver the vehicle is about to do. DOF prioritization has allowed us to develop a modular, easily applicable and extendable motion control architecture for U-CAT vehicle, which is meant for archaeological shipwreck penetration. We demonstrate the benefits of this architecture by developing an remotely operated vehicle autopilot for depth and yaw using a nonlinear state feedback controller. We also show the extensibility of the approach by controlling 3 DOFs of a fully autonomous U-CAT. The real-time experimental results show high position tracking precision (depth RMS error: 1.9 cm; yaw RMS error: 2.5°) Comparative experiments justify the use of DOF prioritization.

Underwater robots conventionally use vision and sonar sensors for perception purposes, but recently bio-inspired sensors that can sense flow have been developed. In literature, flow sensing has been shown to provide useful information about an underwater object and its surroundings. In the light of this, we develop an underwater landmark recognition technique which is based on the extraction and comparison of compact flow features. The proposed features are based on frequency spectrum of a pressure signal acquired by a piezo-resistive sensor. We report experiments in semi-natural (human-made flume with obstacles) and natural (river) underwater conditions where the proposed technique successfully recognizes previously visited locations.

This paper describes an underwater robot navigation strategy in flow. Our aim is to demonstrate that knowing the relative flow speed is advantageous because it permits using more energy efficient and stable control for trajectory following. We use a biomimetic robot that moves in uniform flow using a side-slipping maneuver. Side-slipping permits the robot to move laterally with respect to the incoming flow by exploiting its passive dynamics. The side-slipping maneuver is controlled by adjusting the heading of the robot with respect to the flow. We implement simple PID controllers for controlling the motion of the side-slipping robot laterally and transversely. Also, we compare the performance of the robot in the case where the robot does not know the flow speed. In this latter case the robot's heading towards the waypoint is controlled and the flow effect is considered as a disturbance compensated by the control algorithm. Comparative experiments demonstrate that it is advantageous for a robot to know not just its speed and orientation with respect to the world's frame of reference but also its local flow-relative speed. It permits the robot to follow trajectories more stable and using less energy. In the discussion section we propose possible future directions for implementing the on board flow-relative control.

This paper studies the modelling, design and fabrication of a bio-inspired fish-like robot propelled by a compliant body. The key to the design is the use of a single motor to actuate the compliant body and to generate thrust. The robot has the same geometrical properties of a subcarangiform swimmer with the same length. The design is based on rigid head and fin linked together with a compliant body. The flexible part is modelled as a non-uniform cantilever beam actuated by a concentrated moment. The dynamics of the compliant body are studied and a relationship between the applied moment and the resulting motion is derived. A prototype that implements the proposed approach is built. Experiments on the prototype are done to identify the model parameters and to validate the theoretical modelling.

Underwater vehicles do not localise or navigate with respect to the flow, an ability needed for many underwater tasks. In this paper we implement rheotaxis behaviour in a fish robot, a behaviour common to many aquatic species. We use two pressure sensors on the head of the robot to identify the pressure differences on the left and right side and control the heading of the fish robot by turning a servo-motor actuated tail. The controller is inspired by the Braitenberg vehicle 2b, a simple biological model of tropotaxis, that has been used in many robotic applications. The experiments, conducted in a flow pipe with a uniform flow, show that the robot is able to orient itself, and keep the orientation, to the incoming current. Our results demonstrate that guidance of a fish robot relative to a flow can be implemented as a simple rheotaxis behaviour using two sensors and a Braitenberg 2b controller.

Turbulent flows are often treated as a noisy environment by control algorithms of underwater robots. However, aquatic animals such as fish have learned to take advantage of certain unsteady flow. Periodic complex flow, such as that found in the wake of cylinders has been shown to offer energy saving opportunities to fish. We built a fish-like robot with an integrated pressure sensor array housed in the head. The robot can control its tail beat synchronization with respect to the periodic oscillations in the flow behind a cylinder. We show that vortices, represented here by pressure maxima, can be detected and exploited to increase the swimming efficiency of the robot fish while it remains rigidly mounted to a force plate. Force measurements show an efficiency gain of 23% when the tail beat of the robotic fish is synchronized at a particular phase lag.

This paper describes a sensing-actuation coupling of a robotic trout that detects changes of the laminar flow speed using an on-board pressure sensor and adjusts its tail-beat frequency for steady swimming. The caudal fin actuator closely mimics the morphology of a real trout, in particular the geometry, stiffness and stiffness distribution of the body and the caudal fin. We hypothesize that the linear relationship between the tail-beat frequency and speed, well-known and proven to hold for all fish studied so far, also holds for an artificial fish. We validate the hypothesis and use the results to derive a linear control law to adjust the tail-beat frequency to the swimming speed. We use an onboard pressure sensor to detect the flow speed and test the actuation in a controlled hydrodynamic environment in a flow pipe.

This paper presents a practical approach to identification of reverse-action pairs for reversibility-based learning to develop safe behaviors. The approach is to analyze reversibility of consecutive actions by introducing metrics defined on the actions and the states. The experiment is conducted in the Player/Stage simulator to test applicability of the approach. The robot is allowed to move freely in the environment by executing two-dimensional actions - moving forward/backward or rotating left/right. Reversibility of a sum of every consecutive pair of actions is then analyzed to identify the ones that undo each other. As a result, a set of action pairs are identified with a general rule “if moved/rotated by X, move/rotate by -X to undo”.

Within the broader scope of underwater biomimetics, in this paper we address the relevance of factors such as shape and elasticity distribution in the ability of a compliant device to imitate the kinematic behaviour of a fish. We assess the viability of myometry as a tool to determine candidate mechanical parameters without relying solely on analytical models; we show that we can obtain elasticity distributions that are both consistent with previous theoretical investigations and experimentally better adherent to the passive kinematics of a biological embodiment (rainbow trout).

Soft actuation materials, such as Ionic Polymer-Metal Composites (IPMCs), are gaining increasing interest in robotic applications since they lead to compact and biomimetic designs. In this paper, we propose the use of soft actuation materials as active joints for propelling biomimetic robotic fish. An analytical model is developed to compute the thrust force generated by a two-link tail and the resulting moments in the active joints. The computed joint moments can be combined with internal dynamics of actuation materials to provide realistic kinematic constraints for the joints. Computational fluid dynamics (CFD) modeling is also adopted to examine the flow field, the produced thrust, and the bending moments in joints for the two-link tail. Good agreement is achieved between the analytical modeling and the CFD modeling, which points to a promising two-tier framework for the understanding and optimization of robotic fish with a multi-link tail. We also show that, comparing to a one-link bending tail, a two-link tail is able to produce much higher thrust and more versatile maneuvers, such as backward swimming.

IPMCs are electroactive materials that bend in electric field. This paper describes a linked manipulator using IPMC joints. We argue that this design reduces the control complexity of an IPMC manipulator and increases the precision of the device. The design rationale stems from our theoretical work in material modeling. It suggests that when electrically decoupled short IPMC strips are connected to rigid links, the control of the IPMC manipulator, which is currently vaguely understood and highly non-linear, can be reduced to simple inverse kinematics serial chain manipulator control without the loss in efficiency. We validate our design by comparing the prototype device to a simple IPMC manipulator commonly investigated in the literature. The results show increased precision, reaction time and reachable workspace. We suggest that such a manipulator is suitable for soft and micromanipulation.

We argue that an ability to determine the reversibility of actions allows a robot to identify safe behaviors autonomously. We introduce a notion of reversibility model and give a definition of model refinement. We implement this on a real robot and observe that, when a reversibility model is refined by the addition of proximity sensors, obstacle avoidance emerges as a side-effect of avoiding irreversible actions. We interpret this as evidence of a deep connection between reversibility and safe behaviour. We also observe that, on the real robot, reversibilities are learned as efficiently as a dedicated reward function. We conclude that reversibility identification may provide an abstract and yet practical method of generating a variety of safe behaviours.

Ionomeric polymer-metal composites (IPMC) are electroactive polymer (EAP) materials that bend when electrically stimulated. As IPMC is a relatively new material, proper control methods have not yet evolved. In this paper the usability of IMPC actuators in real world applications is examined from the point of view of precision control. We propose a classical inverted pendulum control problem as a testbed. We suggest that if the pendulum can be balanced, then we have proven the usability of IPMC actuators for precise control tasks. In this paper we describe an inverted pendulum system driven by an IPMC actuator with PC-based control and a camera in the feedback loop. We report the preliminary experimental results in controlling the pendulum and discuss further improvements. To our knowledge this is the first attempt to control a system or manipulate an object with IMPC actuators in a feedback loop