Submarine ML Navigation Control Under Communicative Isolation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The navigation of submarines and submersibles is complex due to the need to interpret various environmental parameters such as pressure, temperature, salinity, and currents, which can change unpredictably, making it challenging to control these watercraft effectively without external data sources.

Innovation Solution

The use of onboard machine learning algorithms to determine relationships between environmental parameters and corresponding actions of the watercraft, allowing for autonomous control and navigation towards targets without reliance on external data or communication.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional navigation techniques (dead reckoning, inertial navigation, bottom contour navigation) are used below periscope depth, then the submarine can navigate without external data sources, but the navigation becomes complex and less reliable due to unpredictable environmental parameter changes

Engineering Contradiction:
Improvenavigation reliabilityVSAvoidcontrol complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The submarine uses its own operational data (speed, course, depth, pitch, roll, yaw) collected during actual voyages to train the machine learning algorithm. This self-service approach eliminates the need for external data sources while improving navigation reliability through adaptive learning from real operational experiences.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

Traditional mechanical navigation systems (dead reckoning, inertial navigation) are replaced with a machine learning-based autonomous control system. The ML algorithm processes environmental parameters and submarine state data to automatically determine navigation actions, reducing control complexity while improving reliability through adaptive decision-making.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Adaptability or versatility

If machine learning algorithms are used for autonomous control, then navigation adaptability to changing environmental conditions improves, but the system complexity and training data requirements increase

Engineering Contradiction:
Improvenavigation adaptabilityVSAvoidsystem complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The machine learning algorithm serves multiple functions: it processes various environmental parameters (pressure, temperature, salinity, density, tide, current), integrates submarine state data (speed, course, depth, pitch, roll, yaw), and autonomously determines navigation actions. This multi-functionality achieves high navigation adaptability while managing system complexity through a unified ML framework.

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Measurement precision

If communication with external sources is maintained for navigation data, then navigation accuracy improves, but the submarine loses communicative isolation and operational autonomy

Engineering Contradiction:
Improveposition determination accuracyVSAvoidoperational autonomy
Core Design Contradiction:
Measurement precisionVSExtent of automation

Solution Approach 1:

The submarine achieves high-position determination accuracy through self-service by using its own operational data collected during voyages to train the ML algorithm. This eliminates the need for external communication while maintaining operational autonomy, as the system learns from and adapts to its own operational experiences.

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP4348371B1Method and apparatus
Publication Date: 2025.05.07 BAE SYSTEMS PLC
  • EP4348371B1 patent drawingFigure 1~2
  • EP4348371B1 patent drawingFigure 3~4
  • EP4348371B1 patent drawingFigure 5

AI summary

A method of training a machine learning, ML, algorithm to control a watercraft is described. The watercraft is a submarine or a submersible submerged in water. The method is implemented, at least in part, by a computer, comprising a processor and a memory, aboard the watercraft. The method comprises: obtaining training data including respective sets of environmental parameters and corresponding actions of a set of communicatively isolated watercraft, including a first watercraft; and training the ML algorithm comprising determining relationships between the respective sets of environmental parameters and the corresponding actions of the watercraft of the set thereof. A method of controlling a watercraft by a trained ML algorithm is also described.