Active SLAM Algorithm for Autonomous Underwater Vehicle Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Underwater inspection missions for complex environments like oil and gas drill centers face challenges due to clutter and obstacles, making pre-planned scripted waypoints unreliable for real-time mapping and inspection, and existing autonomous exploration algorithms fail to account for growing localization uncertainty.
Innovation Solution
An active SLAM algorithm, specifically the Expectation-Maximization Exploration algorithm, is applied to enable an autonomous underwater vehicle to autonomously generate accurate maps by estimating virtual landmarks and managing localization uncertainty, using sensors like multibeam sonars, lidars, and cameras, and fiducial markers for localization and mapping.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If pre-planned scripted waypoints are used for navigation, then the inspection path is predetermined and easy to execute, but the system cannot adapt to clutter and obstacles in real-time
Solution Approach 1:
The navigation system transitions from static pre-planned waypoints to dynamic active SLAM that continuously adapts to environmental features. The system dynamically constructs maps and adjusts navigation paths in real-time based on sensor data and detected landmarks, enabling adaptation to clutter and obstacles while maintaining manageable complexity through algorithmic approaches.
Solution Approach 2:
The AUV performs self-localization and self-mapping without external guidance. The active SLAM algorithm enables the vehicle to autonomously detect virtual landmarks, update its own position estimate, and reconstruct environmental features independently, eliminating the need for complex external positioning infrastructure.
2Measurement precision
If traditional SLAM algorithms are used, then mapping can be performed, but localization uncertainty grows over time reducing accuracy
Solution Approach 1:
The active SLAM algorithm implements continuous feedback by repeatedly detecting and re-localizing on virtual landmarks throughout the mission. This feedback mechanism corrects drift and prevents localization uncertainty from accumulating over time, maintaining accuracy throughout extended inspection missions.
Solution Approach 2:
The system pre-constructs virtual landmarks and their associated uncertainty estimates before actual navigation begins. By having these reference points prepared in advance with known characteristics, the system can quickly re-localize when encountering them, preventing error accumulation without requiring continuous complex computations.
3Manufacturing precision
If dense sensor data is collected for accurate mapping, then map quality improves, but processing time and computational load increase
Solution Approach 1:
The system extracts only the essential features needed for navigation and mapping - virtual landmarks with key geometric properties - from the dense sensor data. By focusing computational resources on identifying and tracking these critical features rather than processing all raw sensor data, the system achieves accurate maps with reduced processing time.
Solution Approach 2:
The active SLAM algorithm changes the parameter representation from raw dense point clouds to simplified virtual landmark models with defined geometric properties and uncertainty estimates. This parameter transformation maintains mapping accuracy while dramatically reducing computational complexity and processing time for real-time operation.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The solution allows for detailed 3D mapping of complex underwater environments, reducing the risk of vehicle entanglement and improving navigation, while ensuring map accuracy and minimizing localization errors, even in feature-sparse regions.
Implementation Method 1
The sensor configuration may include one or more multibeam sonars
Implementation Method 2
The sensor configuration may include one or more multibeam sonars, lidars, and cameras
Data Source
AI summary
Embodiments of the present disclosure are directed towards a system and method for performing an inspection of an underwater environment. Embodiments may include providing an autonomous underwater vehicle (“AUV”) and performing an inspection of an underwater environment using the AUV. Embodiments may further include acquiring real-time sensor data during the inspection of the underwater environment and applying an active simultaneous localization and mapping (“SLAM”) algorithm during the inspection, wherein applying includes estimating one or more virtual landmarks based upon, at least in part, at least one past measurement and a current estimate of AUV activity.