Source-linked AI summary
Acoustic-based Guidance for Automatic Docking of Holonomic AUVs
Ravi Regalo, David Cabecinhas, António Pascoal
TL;DR
Battery constraints make persistent AUV operations dependent on reliable docking, yet fully acoustic docking for fully actuated vehicles has not been demonstrated in the cited prior work. This paper combines two USBL sensors with relative-pose estimation and geometric trajectory tracking; the architecture was validated in simulation and experimentally demonstrated on Medusa. The trials included multiple consecutive successful docking manoeuvres, although confined-water multipath restricted reliable communication to a small aligned region.
Problem
The paper addresses the lack of demonstrated docking for fully actuated AUVs that relies solely on acoustics for both mid-range and terminal guidance, a relevant capability for docking stations that recharge batteries and offload data.
Source: §1 · p. 1Method
Results
Takeaways & Limitations
The demonstrations support the feasibility of fully acoustic docking with the Medusa platform, including operation in low-visibility conditions where optical systems fail.
Source: §7 · p. 6In the confined water tank, multipath limited reliable communication to a small region where the vehicle USBL was aligned with the docking-station boresight.
Source: §6.2 · p. 5
Abstract
from arXiv · showhide
This paper describes a system to automatically dock an AUV onto a docking station without precise knowledge of the position and orientation of the latter, in the presence of unknown ocean currents, using a fully acoustic sensing architecture. The system relies on a pair of Ultrashort Baseline sensors, one onboard the vehicle and one installed on a seabed-resident docking station, enabling operation in low-visibility environments where cameras are ineffective. Relative orientation is estimated by a nonlinear complementary filter on $SO(3)$, while an Extended Kalman Filter provides relative position, supplying pose estimates to a geometric controller on $SE(3)$ that executes the docking manoeuvre. The complete system is implemented in a dedicated software suite and validated in simulation and water trials.
1. INTRODUCTION
Battery limits constrain AUV endurance, motivating docking stations for periodic recharging and data offloading. This paper addresses the lack of demonstrated fully acoustic docking for fully actuated AUVs with a dual-USBL system and reports successful simulation and water-trial demonstrations.
1. INTRODUCTION
- Docking stations can support longer-term seabed operations by enabling periodic underwater battery recharging and data offloading.The introduction also cites reduced logistics costs, improved safety through infrastructure inspection, and persistent environmental monitoring as potential advantages.Source: §1 · p. 1
- The paper targets the gap in fully actuated AUV docking that uses acoustic sensors for both mid-range and terminal guidance.Prior fully actuated examples still used cameras for terminal guidance, while acoustic information was absent or limited to initial approach.Source: §1 · p. 1
- The proposed system uses two USBL sensors, one on the AUV and one on the docking station, for fully relative navigation without precise prior docking-station pose knowledge.An acoustic link gives the vehicle access to docking-station USBL measurements; the system also compensates for unknown ocean currents.Source: §1 · p. 1
- Multiple consecutive successful docking manoeuvres were demonstrated experimentally under challenging environmental conditions, alongside realistic simulation evaluation.The system was implemented in ROS and tested on the Medusa AUV with a custom docking station.Source: §1 · p. 2
2. SYSTEM OVERVIEW AND VEHICLE MODELLING
The section establishes the vehicle and docking-station frames, summarizes the Medusa platform and its dynamics, and describes the paired acoustic measurements used for relative navigation.
2. SYSTEM OVERVIEW AND VEHICLE MODELLING
- The inertial frame I uses North-East-Down convention, the body-frame x-axis follows the vehicle bow, and the docking-station frame D coincides with B when docked.The notation defines η as the position and orientation of B in I, ν as body-expressed linear and angular velocities, and τ as applied forces and moments.Source: §2 · p. 2, Figure 1 · p. 2
- The vehicle dynamics are written using rigid-body and hydrodynamic mass, Coriolis and centripetal effects, damping, restoring forces and moments, and external disturbances.The passage identifies M, C_RB, D, g, and τ_d with those respective dynamics terms.Source: §2 · p. 2
- The Medusa AUV is fully actuated in the horizontal plane and carries GNSS, DVL, depth, AHRS, and acoustic modem/USBL sensors; the docking station also carries a modem and USBL.The docking-station USBL points toward its opening, while the vehicle's USBL and modem point upward; directional beams create blind zones above and behind the station.Source: §2 · p. 2, Figure 2 · p. 2, Figure 3 · p. 2
- The two USBL systems provide reciprocal relative-position measurements: the vehicle measures the direction to the station, and the station measures the direction to the vehicle.The station-side measurement reaches the vehicle through acoustic communication, and each direction is represented by bearing and elevation or its unit vector.Source: §2 · p. 2
3. DOCKING STRATEGY
Docking begins with an approach that reaches the region of reliable acoustic communication, then switches to relative navigation and terminal trajectory tracking. The final path is generated online as a safe, smooth reference rather than as a distance-optimal route.
3. DOCKING STRATEGY
- The docking task starts from an arbitrary condition with only approximate inertial station position known and unknown station orientation; terminal navigation requires acoustic contact.The vehicle may begin outside the region where communication with the docking station is feasible.Source: §3 · p. 3
- Before acoustic contact, the AUV uses dead reckoning, absolute fixes, and path following to reach a homing region, then searches circularly at fixed altitude if needed.The transition waits for several consecutive valid acoustic packets to avoid switching prematurely during intermittent packet loss.Source: §3 · p. 3
- After reliable acoustic contact, the vehicle estimates its pose in the docking-station frame from USBL measurements and tracks a terminal docking path.The manoeuvre transitions from the initial approach to fully relative navigation for the final phase.Source: §3 · p. 3
3.1 Docking trajectory generation
- The terminal trajectory is generated once online when acoustic communication and relative-pose estimates are available; it prioritizes a safe, smooth, feasible reference over distance optimization.It is not continuously replanned.Source: §3.1 · p. 3
- The path rotates toward a homing point, advances toward it, adjusts heading to align with the docking-station x-axis, and then enters along that axis.The vertical coordinate converges independently, while jerk-limited reference curves provide continuity in position, velocity, and acceleration.Source: §3.1 · p. 3
4. DUAL-USBL NAVIGATION SYSTEM
The dual-USBL navigation system estimates relative orientation with a complementary filter and relative position with an EKF. It also rejects outliers and fuses delayed acoustic measurements at their measurement times.
4.1 Orientation filter
- The orientation filter follows Mahony et al.’s Explicit Complementary Filter, using line-of-sight and gravity directions to estimate vehicle rotation relative to the docking station.It assumes a horizontal seabed, and the filter is reported to have almost global asymptotic stability.Source: §4.1 · pp. 3–4
4.2 Position filter
- The position filter is a complementary-style EKF in which DVL velocity dominates at higher frequencies and USBL observations provide lower-rate corrections.Its prediction uses vehicle orientation and measured body velocity; the USBLs are mounted at known offsets from the vehicle and docking-station frames.Source: §4.2 · p. 4
4.3 Robust fusion of delayed acoustic measurements
- Innovation gating rejects USBL updates above a chosen chi-squared threshold, while fixed-lag buffering rolls back and re-propagates estimates so delayed measurements are fused at their timestamps.The attitude filter gates a two-dimensional bearing residual projected onto the tangent plane of the predicted unit vector.Source: §4.3 · p. 4
5. TRAJECTORY-TRACKING CONTROLLER
A geometric inverse-dynamics PID controller in SE(3) tracks the terminal docking trajectory. Its acceleration commands use position and attitude tracking laws, with position integral action for slowly varying disturbances.
5. TRAJECTORY-TRACKING CONTROLLER
- The proposed terminal trajectory tracker is a geometric inverse-dynamics PID controller formulated on SE(3).The control law uses the vehicle dynamics model to compute commanded forces and torques; the Medusa model was previously identified and experimentally verified.Source: §5 · p. 5
- Position and attitude tracking errors are defined in the body frame relative to desired inertial references and their corresponding body-frame signals.The controller uses position, velocity, attitude, and angular-velocity errors.Source: §5 · p. 5
- PID-like translational and PD-like rotational laws with acceleration feedforward generate commanded accelerations; integral action appears only in position control to reject slowly varying disturbances such as ocean currents.The position integral state is driven by velocity error and a term proportional to position error.Source: §5 · p. 5
- The commanded translational acceleration combines reference acceleration with velocity and position feedback and position-integral terms.The controller defines diagonal positive-definite gains for the translational and rotational error laws.Source: §5 · p. 5
- Under suitable gains, position error is globally asymptotically stabilized and attitude error almost-globally stabilized, except at a measure-zero set of unstable attitude critical points.The controller’s stability proof is omitted and follows standard Lyapunov arguments.Source: §5 · p. 5
6. RESULTS
The proposed filter and controller successfully track the terminal docking manoeuvre under simulated current, and the real Medusa vehicle completes five consecutive docking trials. The water-tank trials were constrained by acoustic multipath and required favorable initial positions for communication.
6.1 Simulation results
- Under a uniform 0.2 m/s current at −90° relative to the docking station, estimated state closely followed the real state and the planned trajectory was successfully tracked.The simulated manoeuvre succeeded.Source: §6.1 · p. 5, Figure 8 · p. 5
6.2 Real-world trials
- The real-world trials used a 5-by-3 m water tank with a depth of 4 m at IST Taguspark.The testing site, Medusa vehicle, and docking station are shown in Figure 9.Source: §6.2 · p. 5, Figure 9 · p. 6
- Tank-wall and surface reflections caused severe multipath, limiting reliable acoustic communication to a small region where the vehicle USBL aligned with the docking-station boresight.Each run therefore had to begin from a favorable position to establish communication and prevent filter divergence.Source: §6.2 · pp. 5–6
- Five consecutive real-world docking trials succeeded after the vehicle started as far from the docking station as possible with intentional misalignment.The reference trajectory was accurately generated and tracked; the trajectory origin was shifted upward to improve acoustic coverage.Source: §6.2 · p. 6, Figure 10 · p. 6
7. CONCLUSIONS
The paper presents a fully acoustic underwater docking solution combining relative navigation with geometric trajectory tracking. It was validated in high-fidelity simulation and demonstrated experimentally on Medusa; future work targets controller stability and robustness to model uncertainty.
7. CONCLUSIONS
- The architecture combines a nonlinear complementary filter for attitude with an EKF for relative position, while a geometric trajectory-tracking controller guides the vehicle into the dock.The system is fully acoustic and addresses docking without reliance on camera-based terminal guidance.Source: §7 · p. 6
- High-fidelity simulations and Medusa experiments confirmed the feasibility and robustness of the proposed docking solution.Source: §7 · p. 6
- Future work will strengthen tracking-controller stability properties and improve robustness to model uncertainty.Source: §7 · p. 6