Doppler Velocity Log

• Also called: DVL, Doppler log
• Category: Sensing › Exteroceptive
• Tags: underwater-robotics, dead-reckoning, acoustic-sensing, state-estimation, sensor-fusion

An acoustic sensor for underwater vehicles that measures velocity relative to the seabed, or to the water, from the Doppler shift of sound echoes, and is the main input to underwater dead reckoning.

Definition

A Doppler velocity log (DVL) is an acoustic sensor that measures how fast an underwater vehicle is moving. It sends pulses of sound along several beams, usually four, angled down and away from the vehicle, and measures the Doppler frequency shift of the echoes returned from the seabed. Each beam gives the velocity component along its own direction; with at least three non-coplanar beams, the three-dimensional velocity of the vehicle in the sensor frame can be solved for, and a fourth beam adds redundancy and an error check. Most DVLs also report the range to the bottom along each beam, which gives altitude above the seabed. The word log is the nautical term for an instrument that measures a ship's speed.

When the seabed is within range, the DVL is in bottom-track or bottom-lock mode and measures velocity over ground. When the bottom is too far away, some DVLs can water-track, measuring velocity relative to a layer of water, which includes the effect of currents. Range and precision depend on the acoustic frequency: lower frequencies reach farther, higher frequencies give finer measurements at shorter range. Bottom lock can be lost over steep or soft terrain, very close to the bottom, in aerated water or when the vehicle pitches strongly, and navigation software must handle these gaps.

Since GPS does not work underwater, the DVL is the main source of motion information for most autonomous and remotely operated underwater vehicles. Its body-frame velocity is rotated into the world frame using an attitude and heading estimate from an IMU or inertial navigation system, then integrated to give position. This dead reckoning drifts with distance traveled, with heading error often the largest contributor, so it is corrected with a pressure sensor for depth, acoustic positioning such as USBL or LBL, terrain or visual features, and surfacing for GPS fixes. Accurate results depend on calibrating the mounting rotation and offset between the DVL and the IMU and on using the correct speed of sound.

Why it matters

Underwater robots cannot use GPS and often cannot see far, so velocity over the seabed is one of the few direct measurements of their motion. The quality of a DVL and of its fusion with inertial sensors largely sets how far an underwater vehicle can travel before its position estimate becomes unreliable.

Key applications

• Dead-reckoning navigation for autonomous underwater vehicles
• Station keeping and low-speed control of remotely operated vehicles
• Aiding inertial navigation systems on survey vehicles
• Seabed mapping and inspection where position accuracy matters
• Altitude keeping above the seabed

In practice

An underwater vehicle that must return to sample sites on the seabed fuses its DVL, IMU and pressure sensor in a factor graph. Each DVL bottom-track reading becomes a velocity factor between consecutive states, rotated by the estimated orientation and corrected for the lever arm between the DVL and the vehicle's reference point. Before deployment, the team calibrates the DVL's mounting rotation by driving straight lines and comparing against acoustic position fixes, since a small yaw misalignment produces a steadily growing sideways error. In operation, readings flagged as invalid when bottom lock is lost are dropped, and a contact event, such as the vehicle landing on a known structure, is added as an extra constraint that pulls the drifting estimate back into place.

Common misconceptions

• A DVL gives position — it measures velocity, which must be integrated with an orientation estimate and drifts over time
• DVL velocity is always relative to the seabed — in water-track mode it is relative to the water, so currents enter the measurement
• Any IMU orientation is good enough for DVL dead reckoning — heading errors turn directly into position errors that grow with distance traveled
• Mounting alignment can be ignored — a small rotation between the DVL and IMU frames causes systematic drift

Who should know this

• Marine and underwater robotics engineers
• Navigation and state estimation engineers for AUVs and ROVs
• Ocean science and survey teams using autonomous vehicles
• Researchers working on underwater SLAM and localization

Related technologies

• Inertial Navigation System
• Acoustic Doppler Current Profiler
• USBL Acoustic Positioning
• LBL Acoustic Positioning
• Pressure Depth Sensor
• Attitude and Heading Reference System

Related terms

• IMU
• Odometry
• Factor Graph
• Sensor Fusion
• Extended Kalman Filter
• Extrinsic Calibration

Related projects

• FAST-LIVO2: Fast, direct LiDAR-inertial-visual odometry from HKU-MARS, published in IEEE T-RO
• GTSAM: Factor-graph optimization library powering many modern SLAM backends
• OpenVINS: The reference open platform for visual-inertial state estimation research
• manif: Header-only C++ Lie theory library for robotics state estimation