Energy-optimal path planning of autonomous underwater vehicles using adaptive flow models

AUVs are an important tool for mapping and understanding the ocean. Expensive and time consuming launch and recovery operations coupled with relatively short deployments limit the effectiveness of current AUVs. Energy-optimal path planning utilizing ocean currents is one way to extend the mission duration of AUVs. Modern AUVs can utilize onboard flow sensing abilities to measure ocean currents at their location. In this paper, an adaptive planner utilizing proper orthogonal decomposition (POD) for flow modeling is introduced. The conventional planner uses a fixed number of basis functions or modes, which can vary between missions. The proposed planner adapts the flow model throughout a mission based on measurements gathered during transit, making the algorithm robust over different flow domains. A case study is carried out on a $40\times 40 km$ flow field with an AUV model, where the proposed planner is compared to the average, best, and worst performance of the conventional planner using different numbers of modes. The comparison is made in ten areas of different flow patterns. The proposed model adaptive planner outperforms the average energy consumption of the conventional planner in 8 of the 10 domains. It outperforms the worst energy result of the conventional planner in all 10 domains.