You know how your phone buzzes in your pocket when you're about to miss a turn — just enough vibration to make you glance down, not enough to shove you off the sidewalk? Now imagine that nudge is a water thruster strapped to your torso, and the sidewalk is a coral reef you're drifting toward at depth. That's RADMCS: a wearable robot that uses monocular depth estimation to sense how far a diver is from a subsea structure and then fires small thrusters to physically guide the diver away, like a force-feedback GPS for your body. The committed claim: this is the first wearable robotic system that provides directional haptic guidance to scuba divers for lateral station-keeping, as distinct from propulsive exoskeletons that just make you swim faster. Prior work in underwater wearables has focused on thrust-augmented locomotion — helping divers move more efficiently. RADMCS flips the function: instead of amplifying the diver's intent, it overrides drift by providing a physical cue the diver can feel and respond to. The distinction matters because lateral proximity control near structures like unexploded ordnance or fragile coral is a safety problem, not a speed problem. The architecture is straightforward and intentionally minimal. A monocular camera feeds depth-estimation algorithms to compute distance to nearby structures. When the diver drifts inside a set threshold, submersible thrusters fire at calibrated intensity to create a perceptible force on the diver's body. The key finding is that the threshold for human perception is remarkably low: 10 percent of maximum thruster output is sufficient to redirect a person's movement underwater. That's important because it means the system doesn't need to overpower the diver — it just needs to tap them on the shoulder, so to speak. Integrity is a genuine bright spot for an early-stage robotics paper. The authors ran IRB-approved human-subjects studies with eight scuba divers across two distinct environments: a controlled swimming facility and open ocean. They tested three things: perception thresholds (can divers feel it?), distance maintenance (does the feedback actually keep them in lane?), and form-fit-function (does the hardware survive real conditions?). Eight participants is a small sample, but the fact that they went to open water at all — with IRB approval, not just a lab tank — puts this ahead of most first-generation wearable robotics papers. The ladder question is where things get honest. There is no real SOTA to compare against because nobody has built this category of device before. Propulsive exoskeletons exist (AquaCyborg, underwater jetpacks), and haptic feedback research exists in terrestrial wearables, but the intersection — haptic guidance for lateral underwater station-keeping — is an empty category. That makes the paper genuinely novel but also makes it hard to benchmark. The 10% thrust threshold is the headline number, but without competing systems, we don't know if that's good or merely sufficient. The milestone gap is substantial. Eight divers in controlled and ocean settings is proof of concept, not proof of utility. The next meaningful number is something like 30-50 divers across varied underwater tasks (ordnance inspection, reef survey, infrastructure maintenance) with quantitative collision-avoidance metrics compared to unassisted divers. The system also needs to demonstrate robustness in turbid water and current, where monocular depth estimation degrades. The compact form factor is promising — the authors explicitly flag expandability — but the path from 'diver can feel it' to 'diver trusts it in operational conditions' is long. The obvious experiment the authors didn't run: a closed-loop distance-maintenance trial in the ocean with quantitative structure-proximity metrics. They did distance-maintenance in the pool and form-fit-function in the ocean, but not the combination. Most likely explanation: ocean trials are expensive, logistically brutal, and they wanted the IRB data from controlled conditions first before attempting the full integration. Saving the harder test for the next paper is the pragmatic read.
