Componentizing autonomous underwater vehicles by physical-running algorithms.

Area of Science:

• Robotics
• Cyber-Physical Systems Engineering
• Marine Technology

Background:

• Autonomous underwater vehicles (AUVs) are complex cyber-physical systems requiring systematic engineering approaches.
• Component-based design simplifies AUV development through reusable and composable elements.
• Current development often relies on computational models, which can be time-consuming and costly.

Purpose of the Study:

• To propose an architecture for designing and optimizing AUV components using a 'physical-running' evaluation.
• To demonstrate a method engineering approach for applying the proposed architecture.
• To validate the feasibility and cost-effectiveness of the physical-running evaluation against simulation-based methods.

Main Methods:

• Proposed a novel architecture for AUV component design and performance evaluation.
• Implemented a computing approach using genetic algorithms for automatic control of a testbed.
• Conducted experiments to determine optimal operating modes for an AUV thruster with a flexible propeller through physical-running evaluation.

Main Results:

• Demonstrated the feasibility of designing and assessing physical AUV components directly with genetic algorithms in real-world settings.
• Eliminated the need for computational models and associated engineering stages for optimization and testing.
• A cost-based model showed that the physical-running approach offers extensive feasibility zones and is more cost-effective than simulation.

Conclusions:

• Direct physical-running evaluation using genetic algorithms is a viable and efficient method for AUV component design and optimization.
• This approach streamlines the engineering process, reducing reliance on computational modeling.
• The physical-running perspective provides a more cost-effective and broadly applicable design strategy for AUVs.