BIND-USBL Lets Surface Drone Fleets Fix Underwater Robots' Navigation Drift
A cooperative localization framework uses fleets of surface vessels as acoustic anchors to prevent underwater robots from drifting during long GPS-denied missions.
ProTakeaways
- BIND-USBL uses a fleet of surface drones as coordinated acoustic anchors to bound underwater vehicle drift in GPS-denied missions.
- Combines N-gon formation geometry, conflict-graph TDMA scheduling, and delayed complementary-filter fusion of USBL fixes.
- Multi-ASV fused fixes reduce standard deviation by 1 over square root of K listening anchors.
- Three ASVs cut cross-track error from ~5 m to ~1 m in 140 m surveys versus a single anchor.
- Formation angle alone can produce 8x error reduction in two-ASV deployments with identical hardware.
- Evaluated in HoloOcean simulator; field validation and public code release remain open.
Underwater robots have a memory problem. Once they dive below the surface, GPS disappears, and they must estimate their position from inertial sensors alone. Small biases compound into wildly incorrect position estimates within minutes, a serious issue for a vehicle expected to survey a seafloor grid or rendezvous with a recovery ship. A new paper introduces BIND-USBL, a framework that treats this as a fleet-level problem: rather than improving the underwater vehicle's own sensors, it uses a coordinated team of surface drones to feed intermittent position fixes down through the water.
Why dead reckoning breaks down
Without external corrections, an AUV's position estimate diverges from truth due to sensor bias and process noise, with error growing unboundedly over time. The usual patch is Ultra-Short Baseline (USBL) acoustic positioning, where a surface vessel pings the underwater vehicle with sound, measures round-trip time and direction of arrival, and combines that with its own GPS position to compute an absolute fix for the AUV.
USBL only works when the AUV is within acoustic range of a surface ship. The authors argue that long-duration navigation failure is driven by the temporal sparsity of fixes rather than the accuracy of any individual measurement: as inter-fix intervals grow, dead-reckoning error diverges. A single support vessel becomes a bottleneck when the survey area is large or the fleet is big.
Surface drones as acoustic anchors
BIND-USBL replaces the crewed ship with a fleet of Autonomous Surface Vessels arranged in a geometric formation around the survey area. The system runs a two-band acoustic protocol: a short-range high-frequency uplink channel on which AUVs ping requesting fixes, and a longer-range medium-frequency downlink channel on which ASVs broadcast the accumulated fix payload. HF scheduling is managed by a TDMA protocol, whereas MF uses an event-triggered priority assignment.