A modular hardware-software architecture of an autonomous underwater vehicle for deep sea exploration

Abstract:

This paper presents the initial development of a hardware-software modular and scalable architecture based on low cost FPGA and ARM processor development boards to implement an Inertial Guidance System, Computer Vision, Stochastic Optimization and Deep Neural Networks for a man portable AUV designed to enable operations to water depths as great as 4000 m. The software is coded by VHDL language running on an FPGA and C/C++ scripts running on an Embedded System. The FPGA and ARM processor are contained in the same chip. The main purpose of the hardwaresoftware architecture is perform some complex tasks of a ROV with human operators like identify sites of scientific interest and make parking strategies to collect underwater samples. The sites of scientific interest could be a new hydrothermal vent or an unknown shipwreck. Also the mission can be reconfigured onboard according to the relevant of the acquired data through the vehicle's sensors. Results from laboratory and AUV sea trials are shown.

Año de publicación:

2017

Keywords:

• Fpga
• Autonomous Underwater Vehicle (AUV)
• Deep Neural Networks (DNN)
• stochastic optimization
• Embedded System

Fuente:

scopus

google