A Review of Underwater Robot Localization in Confined Spaces

Haoyu Wu, Yinglong Chen, Qiming Yang, Bo Yan, Xinyu Yang

TL;DR

Navigating underwater robots in confined spaces like caves or shipwrecks is a major challenge due to poor visibility and lack of features. This review paper surveys current autonomous localization methods, primarily SINS/DR and SLAM, and discusses their suitability and limitations for these demanding environments, highlighting future research directions.

What Problem Does This Paper Address?

Exploring underwater environments is crucial for resource discovery, scientific research, and infrastructure inspection. However, confined spaces present unique difficulties for underwater robots. Unlike open water, these environments often suffer from low light, high turbidity, and a scarcity of distinct features, rendering traditional localization systems (like GPS or acoustic beacons) ineffective or prone to significant errors. This makes it challenging for robots to know their precise location, hindering their ability to perform tasks autonomously and safely. The paper aims to consolidate and review the various methods developed to overcome these specific challenges.

Methods

The paper categorizes underwater robot localization methods into three fundamental approaches:

  • SINS (Inertial Navigation System) / DR (Dead Reckoning) Localization: This relies on integrating data from inertial measurement units (IMUs) and other sensors like Doppler Velocity Logs (DVLs) and compasses. While autonomous, SINS/DR methods are prone to cumulative errors over time, especially in confined spaces. The paper discusses both single-method applications and, more importantly, integrated approaches like loosely and tightly coupled SINS/DVL fusion filtering (e.g., using Kalman filters) to mitigate these errors.
  • SLAM (Simultaneous Localization and Mapping): This approach involves building a map of the environment while simultaneously localizing the robot within that map. The paper details various SLAM types, including Visual SLAM, Sonar SLAM, and Laser SLAM, each suited to different environmental conditions and sensor capabilities. It highlights the challenges of feature extraction and loop closure in turbid underwater environments and discusses image restoration/enhancement techniques for visual SLAM.
  • Multi-Sensor Fusion SLAM: Recognizing the limitations of individual methods, this approach combines data from multiple sensors (e.g., vision, sonar, laser, IMU, DVL) to achieve more robust and accurate localization. The paper emphasizes the importance of fusion filtering algorithms to effectively integrate these diverse data streams.

Additionally, the paper touches upon "Other Localization Methods," including fuzzy localization, polarized light navigation, and bio-inspired approaches mimicking fish lateral lines, which offer unique advantages in specific scenarios.

Datasets

Not discussed in the paper. The paper is a review and does not present new experimental results or datasets.

Key Insights and Ideas

  • Confined spaces are a distinct challenge: The paper strongly emphasizes that localization in confined underwater spaces is fundamentally different and more difficult than in open water, requiring specialized approaches.
  • No single silver bullet: The review clearly indicates that no single localization method is universally superior. The choice of method depends heavily on the specific environmental characteristics (turbidity, features, lighting) and available sensor payload.
  • Fusion is key for robustness: The trend towards multi-sensor fusion, particularly combining SINS/DR with SLAM, is a critical takeaway. This integration leverages the complementary strengths of different sensor types to overcome individual weaknesses.
  • Bio-inspired methods are emerging: The inclusion of bio-inspired localization techniques highlights a promising avenue for future research, drawing inspiration from nature's solutions to underwater navigation.
  • Sensor limitations drive method development: The paper implicitly shows how the limitations of sensors (e.g., acoustic interference, visual dependence on light) have driven the development of sophisticated algorithms and fusion strategies.

Key Results

As a review paper, it doesn't present new quantitative results. However, it synthesizes findings from numerous studies:

  • SLAM's advantage: Studies cited show SLAM positioning achieving accuracy levels significantly higher (e.g., 0.5 orders of magnitude) than pure DR positioning in confined spaces.
  • Fusion improves accuracy: Integration of SINS and DVL is shown to significantly reduce positioning errors compared to using either method alone.
  • Sonar's robustness: Sonar SLAM is highlighted as being less affected by low light and turbidity compared to visual SLAM, making it a common choice for challenging underwater conditions.
  • Visual SLAM challenges: Visual SLAM's performance is heavily influenced by illumination and turbidity, requiring image enhancement or restoration techniques.
  • Laser SLAM potential: Laser SLAM offers higher imaging accuracy than sonar and a longer operating range than visual SLAM, though it's affected by water scattering.

Limitations and Future Work

  • Sensor limitations: The paper notes the ongoing challenge of developing high-precision sensors specifically for confined underwater environments. Current bio-inspired sensors, while promising, are not yet fully capable of high-precision tasks.
  • Environmental variability: The dynamic and unpredictable nature of confined underwater spaces remains a significant hurdle for all localization methods.
  • Computational cost: Advanced fusion and SLAM algorithms can be computationally intensive, posing a challenge for resource-constrained underwater robots.
  • Future Directions:
    • Multi-sensor fusion: Continued research into optimal sensor combinations and advanced fusion filtering algorithms (including adaptive AI-based methods) is crucial.
  • Novel sensors: Development of new, high-precision sensors is a critical area for future work.
  • Bio-inspired localization: Further exploration and refinement of bio-inspired navigation methods hold significant promise.
  • Dynamic target detection: Incorporating detection of dynamic targets into localization benchmarks can enhance obstacle avoidance and overall accuracy in complex environments.

Summary updated

Abstract

Underwater robots often encounter the influence of confined underwater environments during underwater exploration. These environments include underwater caves, sunken ships, submerged houses, and pipeline structures. Robot positioning in these environments is strongly disturbed, leading not only to the failure of some commonly used positioning methods but also to an increase in errors in positioning systems that normally function well in open water. In order to overcome the limitations of positioning methods in confined underwater environments, researchers have studied different underwater positioning methods and have selected suitable methods for positioning in such environments. These methods can achieve high-precision positioning without relying on assistance from other platforms and are referred to as autonomous positioning methods. Autonomous positioning methods for underwater robots mainly include SINS/DR positioning and SLAM positioning. In addition, in recent years, researchers have developed some bio-inspired autonomous positioning methods. This article introduces applicable robot positioning methods and sensors in confined underwater environments and discusses the research directions of robot positioning methods in such environments.
BibTeX
@article{wu2024review,
  author = {Haoyu Wu and Yinglong Chen and Qiming Yang and Bo Yan and Xinyu Yang},
  title = {{A Review of Underwater Robot Localization in Confined Spaces}},
  journal = {Journal of Marine Science and Engineering},
  year = {2024},
  doi = {10.3390/jmse12030428},
}
Figure