Introduction to Diver Motion Capture for Conduit Reconstruction
Where the Survey Line Loses Three Meters of Passage
A sidemount diver exits a sidewall restriction into a bedding-plane chamber at Peacock Springs. The line goes straight through the narrow slot and out the far side. Three meters of chamber north and south of the line never gets surveyed, because the diver never swims there. The sketch records a tunnel; the cave is actually a room. Line-and-tape surveys carry this bias everywhere the line is narrower than the chamber.
Diver motion capture closes that gap. Every fin-stroke, body roll, and depth adjustment carries geometric information about the diver's envelope in the conduit. A sidemount configuration in a 1-meter slot produces a different motion trace than the same diver in a 3-meter chamber — shoulder roll is constrained, fin amplitude drops, bottom timer readings show deeper bottom-of-sweep positions. The motion trace itself reveals chamber dimensions that the line never touched.
The principle generalizes to every body articulation a cave diver makes. A backmount diver in a vertical restriction has to rotate ninety degrees onto one side; a CCR diver clearing a slot has to manage the loop and counterlung positioning differently from an open-circuit diver. Each of these adjustments is a measurable kinematic event, and each event carries information about the chamber that produced it. A diver who reaches up to clip a stage bottle to a ceiling tie-off reveals the ceiling's height in that location. A diver who switches a regulator at a known depth reveals the local gas-density environment, which feeds into the sound-speed model EchoQuilt uses for acoustic returns. None of this requires the diver to do anything different from normal cave-diving practice. The data is in the dive; the instrument simply records it.
For long-penetration dives the motion archive becomes its own survey product. A six-hour Wakulla Springs push at 80 meters depth produces a six-hour motion record that, when reviewed at the surface, shows the team's pace, the gas-switch decisions, the bottom-timer holds at deco stops, and the drift through every restriction. Survey leaders who debrief from the motion record alone — without the acoustic data — already gain a richer understanding of the dive than the diver's own slate notes can provide.
Advancements in Sensor Fusion for Underwater SLAM (PMC) reviews how IMU, DVL, and vision fusion suppresses drift in underwater SLAM pipelines. MEMS IMU Navigation with Dead-Reckoning (MIT-WHOI) documents IMU dead-reckoning drift and the acoustic updates needed to correct it. Lost in the Deep: Dead Reckoning Performance Evaluation (ACM) provides empirical evaluation of underwater dead-reckoning drift across hardware tiers. The academic consensus: IMU alone drifts, but fused IMU-plus-acoustic stays within survey-grade bounds for dives of typical length.
Stitching Motion Into the Quilt
EchoQuilt treats diver motion as half the survey data stream, equal in weight to the acoustic returns. Each fin-stroke is logged against its accelerometer signature. Each body rotation is logged against gyroscope rates. Depth changes come from pressure. The IMU traces combine into a six-degree-of-freedom body trajectory through the cave, and that trajectory is the spine onto which acoustic patches get stitched.
The quilt metaphor extends naturally: the motion trajectory is the warp, and the acoustic patches are the weft. A well-made quilt needs both threads. Motion without sound gives a diver path with no walls; sound without motion gives walls with no known diver position. Woven together they produce a passage reconstruction with diver location, wall location, and timestamp at every patch on the quilt. SVIn2: multi-sensor fusion underwater SLAM demonstrates a tightly-coupled visual-inertial-sonar-pressure SLAM system that proves the fusion approach works in cave environments.
For divers, the practical translation is that rig configuration matters for motion capture. Backmount and sidemount produce different IMU signatures — sidemount rolls the whole body, backmount holds it stable and swings the fins. The rig configuration tradeoffs writeup walks through how each config affects the reconstruction in concrete numbers.
What survey teams most often miss on a first motion-capture dive is how much the floor matters. Cave floors contain significant information: breakdown piles, sand banks, biomass mounds. A diver who hovers five centimeters above the floor versus one hovering fifty centimeters produces different motion traces against the same chamber. Logging floor-clearance as a routine data field lets EchoQuilt disambiguate chamber heights from diver buoyancy behavior. Biologists working dry caves have been doing the same thing for decades, and the floor logging survey approach for guano-pile surveys without physical disturbance applies the same data-discipline principle to roost work.
Work on autonomous cave exploration shows where the engineering headroom sits. Demonstrating CavePI autonomous cave exploration (arXiv) puts semantic guidance and caveline tracking on an AUV; the diver-carried equivalent inherits the same sensor stack but leaves the navigation decisions to the human. Underwater Cave Mapping using Stereo Vision (NSF) explores the optical portion of that stack in cavern-zone conditions, providing useful context for why pure motion capture works deeper in the cave where stereo fails.
Advanced Tactics for Motion-Capture Surveys
Three practices extract maximum value from diver motion data. First, calibrate the IMU against the rig, not just against the diver. Hardware-mounted IMUs drift relative to the diver's body when the rig shifts — stage-bottle swaps, mask clears, and valve drills all move the IMU in the diver's frame. A post-calibration step after each major rig manipulation produces cleaner trajectories. GUE-trained teams are already running valve drills on every dive; adding a recalibration step costs ten seconds.
Second, mark each survey station with a deliberate motion signature. A controlled 360-degree slow rotation at a station gives EchoQuilt an unambiguous anchor point — the motion signature is distinctive enough that the reconstruction pipeline uses it as a hard registration point for the quilt. Teams that station-mark with rotations cut reconstruction noise by an order of magnitude.
Third, log scooter states in the motion stream. When the survey transitions from swim to scooter, the IMU sees a stepwise change in mean speed, vibration, and body orientation. Annotating the transition explicitly prevents the reconstruction pipeline from misreading the acceleration as diver propulsion. Teams running mixed swim-and-scooter dives on long WKPP pushes find this distinction changes the quality of the resulting quilt. Detailed scooter mapping speeds numbers cover how Suex, Halcyon, and other DPV models compare in motion-trace fidelity for survey-grade work.
A final tactic worth adopting: review motion data after every dive, not just acoustic data. The motion trace shows where divers worked hard — high fin-stroke amplitude, elevated rates of body rotation — and those are almost always the survey-relevant sections. Directing team attention to the hard-working portions of the dive before writing the survey report focuses analysis on the geometrically-interesting parts of the cave. Teams running long Hölloch or Sistema Huautla expeditions report that this single review habit catches more survey-relevant chambers than any other post-dive QC step, because the cave's hardest passages are also its most morphologically rich, and the motion trace is the most reliable cue for where the team encountered them.
Join the Waitlist for Cave Diving Survey Teams
If your survey technique still assumes the line captures the passage, the motion-capture layer adds a lot of chamber that you have been missing. We are rolling hardware first to Florida springs teams, Yucatán survey projects, and select French sump-push and Lot basin teams that run sidemount or backmount configurations on active expedition projects with multi-season mapping goals. Leave your email with a short note on your primary rig config, your scooter use pattern, and the typical penetration distance for your home cave; we prioritize cohort members who can test across configs rather than replicating a single setup. Our field team will follow up within a week to size the hardware kit for your specific expedition profile and target system.