Towards Energy-Aware Feedback Planning for Long-Range Autonomous Underwater Vehicles

Area of Science:

• Robotics
• Oceanography
• Autonomous Systems

Background:

• Ocean ecosystems exhibit complex spatiotemporal variability requiring long-term autonomous underwater vehicle (AUV) data collection.
• New long-range autonomous underwater vehicles (LRAUVs) offer high endurance and energy-aware capabilities for studying oceanic phenomena.
• Uncertain ocean currents significantly impact AUV trajectories, making simple path planning impractical due to potential deviations.

Purpose of the Study:

• To present an energy-aware feedback planning method for LRAUVs operating in uncertain underwater environments.
• To address challenges in state estimation and incorporate state uncertainty into path planning.
• To develop a computationally tractable approach for synthesizing energy-aware feedback plans.

Main Methods:

• Utilized the kinematic model of an LRAUV considering motion and sensor uncertainties.
• Incorporated ocean dynamics from predictive ocean models to analyze water flow patterns.
• Introduced a goal-constrained belief space for computationally tractable feedback plan synthesis.
• Synthesized energy-aware feedback plans using sampling and ocean dynamics for different water current layers.

Main Results:

• Developed a novel energy-aware feedback planning method for LRAUVs.
• Demonstrated the method's ability to generate strategies for navigating uncertain ocean currents.
• Validated the approach through extensive simulations using the Tethys vehicle's kinematic model and real ocean data.

Conclusions:

• The proposed method effectively generates energy-aware feedback plans for LRAUVs in dynamic ocean environments.
• This approach enhances the reliability and efficiency of oceanic data collection by mitigating disturbances from water currents.
• The feedback planning strategy enables LRAUVs to navigate from initial to goal locations while managing uncertainties.