Submarine ML Navigation Under Communicative Isolation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Submarines and submersibles face unique navigation challenges due to the need for self-reliant control in underwater environments, where traditional navigation techniques are unavailable, and environmental parameters such as pressure, temperature, salinity, and currents are complex and unpredictable, making autonomous control difficult.
Innovation Solution
A machine learning algorithm is trained onboard the watercraft using environmental data and control actions to determine relationships between these parameters and appropriate responses, enabling autonomous navigation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional navigation techniques (satellite navigation, RADAR, celestial navigation) are used for submarines, then navigation accuracy is improved, but these techniques become unavailable when submerged below periscope depth
Solution Approach 1:
The submarine performs self-learning by automatically recording environmental parameters (temperature, salinity, pressure, depth, speed, course) and human operator actions during voyages. The system stores this training data onboard and uses it to train machine learning models independently, without requiring external data transmission or intervention, enabling autonomous navigation adaptation
Solution Approach 2:
The system collects and stores training data during preliminary voyages before autonomous operation is needed. By accumulating environmental parameter data and corresponding human actions in advance, the submarine prepares a knowledge base that enables autonomous navigation decisions when traditional navigation methods are unavailable
2Extent of automation
If machine learning algorithms are trained using onboard computers with limited resources, then autonomous control capability is improved, but training time and computational resources are constrained
Solution Approach 1:
The system implements a staged training approach where the machine learning model is first trained on a subset of training data during voyages, then continues training on additional data when computational resources and time become available. This partial training approach allows the system to achieve functional autonomy while continuing to improve through incremental learning
Solution Approach 2:
The machine learning training process operates continuously throughout the submarine's operational life. The system continuously accumulates training data from environmental sensors and operator actions, and continuously trains the model during available computational windows, ensuring the autonomous control capability improves over time without interruption
3Adaptability or versatility
If environmental parameters (temperature, salinity, pressure, currents) are monitored in detail, then navigation adaptability is improved, but system complexity and data processing requirements increase
Solution Approach 1:
The system combines multiple environmental parameter measurements (temperature, salinity, pressure, depth, speed, course) with human operator actions into a unified training dataset. By merging these diverse data types into a single comprehensive knowledge base, the system reduces the complexity of processing separate parameter streams while maintaining full adaptability to environmental variations
Data Source
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.