Towards Physical Underwater Robotic Assistance for Scuba Diver Movement in Confined Spaces

Demetrious T. Kutzke; Junaed Sattar

cs.ROOct 1, 2026

Towards Physical Underwater Robotic Assistance for Scuba Diver Movement in Confined Spaces

Authors: Demetrious T. Kutzke, Junaed Sattar

Abstract

Scuba divers are taught to control their depth to avoid rapid ascents and descents, which could result in serious injuries such as gas embolisms and barotrauma. However, many underwater tasks necessitate lateral control, maintaining distance between subsea structures such as coral reefs, submerged drilling instrumentation, or unexploded ordnance. In this work, we discuss a first-of-its-kind wearable robotic solution providing thruster-actuated directional guidance to a diver, as distinct from prior propulsive-assistance exoskeletons. We introduce ``Robotic Assisted Diver Movement in Confined Spaces'' (RADMCS), a wearable robot that assists divers in maintaining a fixed distance from subsea structures by leveraging perception techniques in monocular depth estimation and force-feedback from submersible thrusters to provide haptic feedback. Its small and compact form factor creates a foundational platform that could be expanded to include more sophisticated control and navigation behaviors. We present results from Institutional Review Board (IRB) in-water studies with eight human scuba diver participants on threshold sensitivity tests in both a closed-water swimming facility and ocean environments; distance-maintaining experiments in a closed-water facility; and form, fit, and function testing in the ocean. We demonstrate that relatively low thrust values (10 percent of maximum) allow robotic direction of a human's movement using the physical sensation of the robot's guidance.

Figures & tables

Figure 1: The Robotic Assisted Diver Movement in Confined Spaces (RADMCS) wearable robot during a field experiment. The diver wears RADMCS on their back, enabling haptic feedback from the robot to the human through a series of thruster movements. If the difference $\delta$ between the setpoint distance $d$ and the actual distance of the diver to the coral reef wall $d^{\prime}$ is greater than zero, the robot indicates that the diver should yaw toward the wall, otherwise yaw away from the wall. The diver responds to the feedback, swimming in the direction indicated by the robot’s thrusters, which creates a pathway to complex navigation and maneuvering for safe navigation along coral walls or confined spaces.

Figure 2: Assembled RADMCS robotic system features with descriptions of parts.

Figure 3: Thruster configurations. The left image column shows the transverse thruster configuration that directs force perpendicular to the robot’s body, in the $\hat{x}$ -direction. The right image column shows the longitudinal thruster configuration that directs force along the length of the robot’s body, in the $\hat{y}$ -direction. The thrusters are labeled and referred to as shown in the figure for all future analysis.

Movement	Longitudinal/Transverse
Turn left (yaw CCW about $\hat{z}$ )	$T1$ reverse and $T2$ forward
Turn right (yaw CW about $\hat{z}$ )	$T1$ forward and $T2$ reverse

Table 1: Directional movement mapping of the haptic feedback from robot to human. In both the longitudinal and transverse thrust configurations the movement instruction is the same.

Event/participant	P1	P2	P3	P4	P5	P6	P7	P8
TS closed-water $225$ mm longitudinal/transverse	$\checkmark$	$\checkmark$
TS closed-water $325$ mm transverse					$\checkmark$		$\checkmark$
Single-sided distance-maintaining	$\checkmark$	$\checkmark$		$\checkmark$	$\checkmark$	$\checkmark$	$\checkmark$	$\checkmark$
TS open-water $325$ mm longitudinal/transverse	$\checkmark$		$\checkmark$
Open-water form, fit, & function								$\checkmark$

Table 2: Participant list and the specific portions of the study they were involved in. Participants are coded from P1–P8. Threshold sensitivity testing is abbreviated as (TS).

Figure 4: Still frames collected during the closed-water TS test for the $225$ mm transverse thruster configuration. During the “ascending” portion of the test, the thruster force-feedback increases in intensity, rotating the diver in the figure.

Panel of Figure 5 — Figure 5: TS for ( $M=2$ ) participants P1 and P2 for CW and CCW rotations in both the $225$ mm longitudinal and transverse thruster configurations in the closed-water environment. The gold solid curve shows the manufacturer-provided thrust–PWM calibration [ 27 ] . The inset illustrates the thruster orientations and resulting force directions used to generate rotational motion.

Figure 6: Simulated distance-maintaining experimental setup in the closed-water swimming facility. A tarp with Kalibr AprilGrids was affixed to the side of the pool. The currents produced from circulation jets created an imperfect surface leading to curvature and distortions in the images. These curvatures along with facets of the camera (including continual autofocus) had downstream effects on the quality of the control algorithm, since the feedback signal is the distance between the robot’s cameras and the wall.

Figure 7: Still frames collected during the open-water ocean TS test for the $325$ mm transverse and longitudinal thruster configurations. During the “descending” portion of the test, the thruster force-feedback decreases in intensity, rotating the diver in the figure until they stop moving.

Figure 8: Simulated distance-maintaining experiment showing the correct turn directions based on distance for participant P2. The top plot shows the PWM curves for $T1$ and $T2$ . The bottom plot shows the distance to the target for both the raw data and the filtered distance data. The red line shows the participant’s end time for the trials, taking $89.89$ s to complete the wall distance. The green region and accompanying aerial perspective show the left turn. The diver responds, turning left. The yellow region shows a right turn, and the diver moves away from the wall.

Figure 9: Simulated distance-maintaining experiment showing the correct turn directions based on distance for participant P2.

Figure 10: Simulated distance-maintaining experiment showing the correct turn directions based on distance for participant P8. The top plot shows the PWM curves for $T1$ and $T2$ . The bottom plot shows the distance to the target for both the raw data and the filtered distance data. The red line shows the participant’s end time for the trials, taking $43.28$ s to complete the wall distance. The green region and accompanying aerial perspective show the left turn. Unfortunately, because the robot lost the distance estimation at approximately $t=12~$ s of run time and never recovered distance, the filtered distance to target control signal never updated, leaving the robot in a high turn left thrust state for the duration of the data run.

Panel of Figure 11 — Figure 11: TS for ( $M=2$ ) participants P1 and P3 for CW and CCW rotations in both the $325$ mm longitudinal and transverse thruster configurations in the ocean environment.

Figure 12: Still frames from the form, fit, and function qualitative test in the ocean environment.

Explore similar work

Sep 20, 2026cs.RO

Towards Reliable Underwater Diver-Robot Interaction: Gesture Design, Interaction Logic, and Real-World Evaluation

Underwater human--robot interaction requires gesture commands that are both easy for divers to use and reliable for robots to recognize. We investigate these aspects through a closed-loop diver--robot interaction framework integrating a compact seven-gesture vocabulary, lightweight landmark-based recognition, and command-level interaction logic. We evaluate the framework through a user study and underwater robot experiments in a laboratory tank and a swimming pool. The user study supported the reproducibility of the gestures after brief learning. Recognition analysis further showed that visual similarity was associated with gesture confusion, while intermediate poses during gesture formation introduced temporal ambiguity. Command-level processing mitigated the effects of transient recognition errors on robot execution, reducing unintended triggers and premature task interruptions. These findings show that reliable underwater gesture interaction depends on human usability, gesture recognizability, and execution reliability in underwater interaction.

Yingqi Liu, Kanzhong Yao, Zimeng Peng +4

University of Science and Technology of China, Hefei, China. · Institute of Artificial Intelligence (TeleAI), China Telecom, Shanghai 200232, China.

Underwater Robots Underwater Human-Robot Interaction Hand Gesture Recognition Co-Speech Gesture Generation Interaction Data+2

Ask PDF

Jun 10, 2026cs.RO

Improving Human Diving Endurance with a Field-Deployable, Untethered Exoskeleton

Human endurance in underwater locomotion is fundamentally restricted by high energetic demands to overcome drag and the finite supply of self-contained breathing gas. While exoskeleton technology can reduce the metabolic cost of humans in terrestrial locomotion, its potential to enhance human endurance during underwater diving remains entirely unexplored. Here, we present DiveMate, a field-deployable, untethered exoskeleton designed to improve human diving endurance via adaptive kick assistance in real-world underwater environments. During naturalistic diving, DiveMate increases the travel distance using a given energy (breathing gas) by 42.9% and extends dive duration by 54.9% through reducing gas consumption rate. Marked reductions in muscle activation indicate a decrease in physiological exertion, with the net gas consumption rate decreasing by 47.0%. Kinematic characteristics and regularity improvements further underpin efficient energy economy. These results suggest that applying exoskeleton assistance is beneficial for improving human diving endurance and augmenting their ability to explore the aquatic world. This study extends the application frontier of exoskeletons and provides a potential reference for the design and assessment of future underwater assistive devices.

Zhihao Zhou, Zhenmeng Ju, Rui Yang +8

Underwater Robots Exoskeleton Assistant

Ask PDF

Jun 10, 2026cs.RO

Semantically-Aware Diver Activity Recognition Framework for Effective Underwater Multi-Human-Robot Collaboration

Effective multi-human-robot collaboration is essential for expanding human-led operations in the challenging and high-risk underwater environment. For autonomous underwater vehicles (AUVs) to become true teammates, they must be able to comprehend their surroundings and recognize a diver's activities to offer assistance and ensure safety. Towards this goal, we introduce DAR-Net, a novel transformer-based framework that analyzes complex underwater scenes to classify diver activities. Our contribution lies in a semantically guided learning formulation that couples transformer-based temporal reasoning with pixel-level scene supervision. This multi-loss training strategy explicitly aligns global activity recognition with local human-robot interaction semantics, which is particularly critical in low-visibility underwater conditions. To address the significant challenge of data scarcity in this domain, we present the first-ever Underwater Diver Activity (UDA) dataset, a foundational resource containing over 2,600 annotated images with pixel-level masks. Through rigorous experimental evaluations in a controlled environment, we demonstrate that DAR-Net achieves promising accuracy in recognizing six distinct diver activities, outperforming state-of-the-art models. While this dataset provides a crucial baseline, our work serves as a pioneering step, laying the groundwork for future research and facilitating the development of more intelligent, collaborative underwater robotic systems.

Sadman Sakib Enan, Junaed Sattar

department of Computer Science & Engineering and the Minnesota Robotics Institute, University of Minnesota, MN, USA

Underwater Robots Autonomous Underwater Vehicles Human Activity Recognition Human-Robot Interaction Collaboration

Ask PDF