On the modification of the SPARUS II AUV for close range imaging survey platform

Yevgeni Gutnik, Aviad Avni, Tali Treibitz, Morel GroperarXiv 2021 · arXiv

High-resolution underwater visual surveys are essential for marine biology and archaeology, but the ocean environment presents significant challenges. Light attenuates rapidly and scatters off particles, requiring cameras to operate at very close range—often within 2 meters of the seafloor—to obtain clear imagery.

To create accurate 3D models, an autonomous underwater vehicle (AUV) must fly precise, overlapping patterns at a constant altitude. However, most torpedo-shaped AUVs are "under-actuated," meaning they cannot move sideways (sway) directly. To compensate for currents, they must "crab" or side-slip, which causes camera tilt and motion blur. Researchers addressed this by modifying a standard SPARUS II AUV into "ALICE," a high-precision hovering platform designed specifically for millimeter-scale seafloor imaging.

Hardware and Control Strategy

The team’s approach combined hardware retrofitting with advanced control theory. They added two lateral tunnel thrusters—one in the nose and one in the tail—to the SPARUS II frame, allowing for independent control of sway and yaw.

The modified SPARUS II AUV (ALICE) with lateral thrusters and custom imaging payload.

Key technical modifications included:

  • Custom Payload: A 9.2-megapixel stereo camera system and 30,000-lumen LED strobes synchronized to fire at up to 10 fps.
  • Thruster Allocation: A Redistributed Pseudo-Inverse (RPI) algorithm managed the over-actuated design, optimizing stability and providing redundancy.
  • Path-Following: An upgraded Line-of-Sight (LOS) controller with a dedicated PID sway controller allowed the AUV to maintain its line without rotating its body, keeping cameras pointed straight down.

Experimental Results

Sea trials in the Mediterranean Sea validated the design. The modified controller reduced the root mean square (RMS) cross-track error from 0.36m to 0.12m—a 65% improvement in accuracy. Stability also improved significantly; roll angles during horizontal turns were slashed from 20°–30° to less than 10°.

A high-resolution 3D reconstruction of a ray, demonstrating the platform's ability to capture fine topographical detail.

The system successfully produced 1mm-resolution 3D models of complex underwater environments, including shipwrecks and the ancient Roman harbor of Caesarea. These results demonstrate that a modified general-purpose AUV can rival dedicated, high-cost imaging platforms.

Future Work

While ALICE is a success, researchers noted that asymmetric vertical dynamics and sensitivity to strong currents remain challenges. Future iterations may include real-time image quality feedback to automatically adjust AUV speed or altitude in response to water turbidity or lighting changes.

Summary updated

Abstract

Based on the need for high resolution underwater visual surveys, this study presents the adaptation of an existing SPARUS II autonomous underwater vehicle (AUV) into an entirely hovering AUV fully capable of performing autonomous, close range imaging survey missions. This paper focuses on the enhancement of the AUVs maneuvering capability (enabling improved maneuvering control), implementation of an state of the art thruster allocation algorithm (allowing optimal thrusters allocation and thrusters redundancy), and the development of an upgraded path following controller to facilitate precise and delicate motions necessary for high resolution imaging missions. To facilitate the vehicles adaptation, a dynamic model is developed. The calibration process of the dynamic model coefficients initially obtained using well accepted formulas, by computational fluid dynamics and in real sea experiments is presented. The in house development of a pressure resistant imaging system is also presented. This system which includes a stereo camera and high power lightning strobes was developed and fitted as a dedicated AUV payload. Finally, the performance of the platform is demonstrated in an actual seabed visual survey mission.
BibTeX
@misc{gutnik2021modification,
  author = {Yevgeni Gutnik and Aviad Avni and Tali Treibitz and Morel Groper},
  title = {{On the modification of the SPARUS II AUV for close range imaging survey platform}},
  year = {2021},
  eprint = {2111.08971},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2111.08971},
}
Figure