Multi-robot coordination for underwater Mothership-Passenger systems

Area of Science:

• Robotics
• Autonomous Systems
• Ocean Engineering

Background:

• Multi-robot coordination faces significant challenges in underwater environments due to limited communication bandwidth.
• Acoustic links in underwater settings restrict centralized planning and hinder decentralized information sharing.

Purpose of the Study:

• To explore a "Mothership-passenger" paradigm for efficient multi-robot coordination under communication constraints.
• To develop and validate algorithms that improve coordination efficiency and robustness in underwater multi-robot systems.

Main Methods:

• Developed two algorithms integrating hierarchical guidance with decentralized planning for multi-robot orienteering.
• Implemented a learned policy on a Mothership vehicle for adaptive robot priority configuration.
• Validated algorithms through simulations and field experiments in an inland lake.

Main Results:

• Demonstrated significant improvements in coordination efficiency and robustness compared to baseline methods.
• Showcased the effectiveness of the "Mothership-passenger" paradigm in broad spatial coverage and reduced mission risk.
• Validated the algorithms' ability to handle stochastic travel costs and mission disruptions.

Conclusions:

• The proposed "Mothership-passenger" paradigm and algorithms effectively address coordination challenges in communication-constrained underwater environments.
• These advancements are crucial for deploying coordinated robotic teams in remote and challenging locations like under-ice ocean regions.
• The study highlights the potential for globally informed guidance to shape local planning under strict constraints.