On the Adaptation of an AUV into a Dedicated Platform for Close Range Imaging Survey Missions

Yevgeni Gutnik, Aviad Avni, Tali Treibitz, Morel GroperJournal of Marine Science and Engineering 2022 · OpenReview

TL;DR

This paper details the successful modification of a standard torpedo-shaped AUV (SPARUS II) into a specialized platform (ALICE) for high-resolution, close-range seabed imaging. By adding lateral thrusters and implementing advanced control algorithms, ALICE achieves significantly improved maneuverability and path-following accuracy, crucial for generating detailed 3D photomosaics.

What Problem Does This Paper Address?

High-resolution visual surveys of the seabed are vital for marine research, archaeology, and infrastructure inspection. However, achieving this requires platforms to operate very close to the seabed, demanding precise motion control, especially in complex environments or when dealing with currents. Traditional torpedo-shaped AUVs, while efficient for long-range surveys, lack the maneuverability for precise, low-altitude, close-range missions. Their underactuated design limits their ability to perform delicate maneuvers, maintain constant altitude and orientation, and accurately follow transects, leading to gaps or overlaps in collected imagery and thus compromising the quality of generated photomosaics. This paper addresses this gap by transforming a capable but limited AUV into a dedicated, highly agile imaging platform.

Methods

The core of this work lies in enhancing the SPARUS II AUV's capabilities through several key modifications:

  • Propulsion System Upgrade: Two lateral thrusters were added to the AUV. These thrusters provide direct control over sway (lateral movement) and decoupled yaw (rotation), significantly improving maneuverability beyond the original surge, heave, and limited yaw control.
  • Advanced Thruster Allocation: A novel, redistributed pseudo-inverse (RPI) method for thruster allocation was implemented. This algorithm optimizes thrust distribution to enhance dynamic stability in roll, particularly at low speeds, and allows for decoupled motion control in yaw and sway, crucial for precise maneuvering. It also dynamically adjusts thruster weights based on measured roll angles to maintain stability.
  • Improved Path-Following Controller: The original Line-of-Sight (LOS) controller was enhanced. The new controller incorporates sway motion control to effectively eliminate cross-track errors and maintain heading alignment with the desired path, even in the presence of lateral disturbances like currents. This is achieved through a switching strategy that prioritizes sway control for small errors and heading control for larger ones.
  • Dedicated Imaging Payload: A stereo imaging system, comprising two 9.2-megapixel cameras and high-power LED strobes, was developed and integrated. This system is designed for high-resolution imaging at close range.
  • Dynamic Modeling and Simulation: A detailed dynamic model of the modified AUV (ALICE) was developed and validated. This model was crucial for designing and optimizing the control system, as well as for performing software-in-the-loop (SIL) simulations to evaluate performance under various conditions before real-world testing.

Datasets

The paper primarily relies on data collected during real-world sea trials conducted off the coast of Israel (Sdot Yam, Achziv, and Haifa Bay). These trials involved:

  • Hydrodynamic Coefficient Identification: Experiments were conducted in calm sea states to measure the vehicle's response to controlled maneuvers (forward motion, vertical ascent/descent, horizontal turns). This data was used to identify the AUV's hydrodynamic coefficients.
  • Control System Calibration and Validation: The SIL simulation environment, incorporating the developed dynamic model, was used to calibrate PID gains for the modified LOS controller. Subsequently, the modified AUV (ALICE) and the original SPARUS II configuration were tested in simulated lawnmower pattern missions and real sea missions to compare their path-following performance.

No new public datasets were introduced; rather, the paper focuses on the collection and analysis of data from its own experimental setups.

Key Insights and Ideas

  • Lateral Thrusters are Key for Close-Range Precision: The addition of lateral thrusters fundamentally transforms an AUV from a long-range surveyor into a precise, agile platform capable of intricate maneuvers required for close-range imaging.
  • Overactuation Enables Sophisticated Control: By making the yaw motion overactuated, the system gains flexibility to optimize for different objectives (e.g., power consumption, stability) through advanced thruster allocation.
  • Integrated Control and Dynamics are Crucial: The success hinges on the tight integration of a realistic dynamic model with advanced control algorithms (thruster allocation and path-following). This allows for fine-tuning and optimization that wouldn't be possible with simpler models.
  • Real-World Validation is Paramount: While simulations are invaluable for development, the paper emphasizes the critical need for real sea experiments to validate the performance of the modified AUV and its control systems under actual environmental conditions.

Key Results

  • Improved Path Following: The modified ALICE configuration demonstrated significantly better path-following accuracy compared to the original SPARUS II. In simulations, cross-track errors were reduced from approximately 0.5 m to less than 1 cm, and heading errors from 20° to 1°.
  • Enhanced Stability: Real sea experiments showed a substantial improvement in roll stability during turns, with roll angles decreasing from 20–30° to less than 10°.
  • Accurate Imaging Transects: ALICE successfully followed imaging transects with a mean cross-track error of approximately 0.12 m in real sea missions, enabling the collection of high-quality, well-overlapped data for photomosaics.
  • Robustness to Disturbances: Simulations indicated that the proposed controller could compensate for significant uncertainties in hydrodynamic coefficients and cross-currents up to 0.25 m/s.

Limitations and Future Work

  • Limited to Specific Missions: The current adaptation is optimized for close-range imaging survey missions. Its applicability to other mission types (e.g., intervention tasks) might require further modifications.
  • Hydrodynamic Coefficient Uncertainty: While the paper addresses uncertainties in hydrodynamic coefficients through simulation, the accuracy of these coefficients still influences performance. Future work could involve more robust online identification methods.
  • Energy Consumption: While the thruster allocation algorithm aims to optimize power, the addition of lateral thrusters likely increases overall energy consumption compared to the original configuration.
  • Future Work: The authors suggest further development could focus on more advanced path-planning algorithms, integration of additional sensors for enhanced situational awareness, and exploring the AUV's capabilities for more complex intervention tasks. The successful demonstration of ALICE's ability to generate high-resolution 3D reconstructions also opens avenues for further research in this area.

Summary updated

BibTeX
@article{gutnik2022adaptation,
  author = {Yevgeni Gutnik and Aviad Avni and Tali Treibitz and Morel Groper},
  title = {{On the Adaptation of an AUV into a Dedicated Platform for Close Range Imaging Survey Missions}},
  journal = {Journal of Marine Science and Engineering},
  year = {2022},
  url = {https://openreview.net/forum?id=XBPtFCcEUL},
}
Figure