Transmission of classified and varying quality underwater maps over constrained networks
Abstract
Autonomous underwater mapping operations are limited by onboard energy constraints in AUVs. The ENDURUNS project proposes to use fuel cells technologies to provide extended duration of AUV mapping operations. This type of mission generates large quantities of multi-beam sonar data available onboard at remote maritime locations. Satellite links can provide connectivity from remote control centers to these unmanned assets, but have limited bandwidth to transfer unprocessed data. This paper describes onboard semantic classification of raw sensor data using deep learning for a compressed representation of data for operators to analyze and selectively demand higher resolution data for specific areas of interest from large data sets. This is aimed at providing access to downstream end users and stakeholders with the required data within the constraints of communication technologies with minimal operational delay.
Event details
- Event
- MetroSea 2019
- Technical Committee
- TC19
- info@metrosea.org
- Place
- Genoa, ITALY
- Time
- 3 October 2019 - 5 October 2019
- Website
- http://www.metrosea.org