Scuba diving has a hard budget: the gas in the tank. A soft robotic exosuit built for finning appears to move that budget by up to 30% — and the number that matters is not speed but how much you breathe.

A focus article in Science Robotics, led by Wu Xinyu at the Shenzhen Institute of Advanced Technology (part of the Chinese Academy of Sciences), surveys the state of underwater exoskeletons. The centrepiece is PEAKED — Performance-Enhancing Assistive Kicking Exosuit for Diving — a soft hip exosuit that assists both hip flexion and extension, the two directions water resists when a diver kicks.

In earlier human experiments, assistance cut respiratory minute volume (RMV, the volume of gas breathed per minute) by 29.77% during flutter kicking, 25.70% during frog kicking and 18.35% during underwater walking. A separate underwater knee exoskeleton produced a 22.7% RMV reduction. Those figures come from different devices and different movement patterns; the paper's own framing is that benefit varies by activity rather than being one universal number.

Why RMV is the right metric: it maps directly onto the finite gas supply, unlike the muscle-activation measures used in earlier swimming exoskeleton work. What it does not do is prove longer safe dives — ventilation is not metabolic cost, and no open-water trials have been reported.

The design problem is the interesting part. On land, an exoskeleton mostly has to carry weight; underwater, buoyancy removes that job and drag takes over, resisting motion in both directions — which is why assisting only one half of the kick cycle fits poorly. Soft structures accommodate the large, multi-plane leg movements of different finning techniques but bring cable-routing entanglement risk; rigid structures transmit torque better but constrain movement, suiting repetitive flutter kicking more than frog kicking or walking.

Practical diving imposes requirements a lab bench does not. The suit has to coexist with tanks, ballast and the rest of the kit, its actuators and electronics have to survive pressure, and it has to come off fast — the paper points to breakaway cables and quick-release buckles in some systems, which matter more than any efficiency claim in an emergency.

Where it goes next: personalised assistance. Current assistance patterns are partly inherited from terrestrial exoskeletons and partly tuned by hand; human-in-the-loop optimisation, where the wearer's own response sets the parameters, is the obvious next step. Asymmetric assistance between the legs could also correct roll imbalance from uneven ballast while preserving thrust.

The honest summary: a credible 18–30% cut in breathing demand depending on how you fin, demonstrated in controlled human tests, with open-water validation, entanglement and emergency egress still unsolved. A promising result, not yet a product.

Sources