Teledyne RESON
Real-time ML in mission-critical sonar
Machine Learning & AIManufacturing & Industrials
Key resultSub-second ML detection inside live sonar systems
Challenge
Sonars produce vast amounts of data with varying noise and signal. Detecting specific signals and patterns is very difficult — and the mission-critical setting demands true real-time processing.
Solution
An ML pipeline of preprocessing, filtering and signal detection, connected inside the sonar system and processing everything in real time.
- Automation. Detection of signals and patterns in vast sonar data streams, fully automated.
- Efficiency. Real-time operation visualising the most relevant information for operators.
- Focus. The system automatically surfaces the signals most likely to matter to users.
Results
- Real-time, sub-second execution
- Relevant signals detected automatically
- Smooth, consistent tracking
Flowtale has contributed with capacity building and establishment of development environments and work processes using cloud solutions for training, tagging and data science tasks — and continues to work with Teledyne RESON on expanding machine learning in mission-critical sonar systems.
Topics
- Machine Learning
- DSP
- DevOps
- Embedded