Assessing Automated and Human Path Planning for the Slocum Glider

Autonomous Underwater Vehicles (AUVs) are a common tool used by oceanographers to study oceans. Most AUVs are operated by pilots who are able to interpret environmental information to make effective mission decisions; this function requires an understanding of the hardware and software of these complex systems. To enable oceanographers and pilots to more easily manage a fleet of gliders, new mechanisms are needed to ease the burden of AUV operation. An automated path planning system is one such tool that could free operators from the tedious task of waypoint selection, and would allow them to focus on scientific and mission critical aspects of managing groups of AUVs. AUV path planning involves selecting a set of waypoints to guide an AUV from a starting location to a destination location while considering obstacles such as shipping lanes, ocean currents, or limited battery resources. Offloading operational tasks to an automatic tool is only feasible if the decisions made by the tool are considered reasonable and can be trusted. We have developed a testbed environment to assess the flight paths and energy consumptions of both an AUV guided by an automated path planning system and human pilots. The testbed environment is based on a new, faster-than-real-time, software only, simulator for the Slocum Glider. In an effort to evaluate this simulator, four pilots with varying backgrounds and glider flight experiences were asked to fly a Slocum glider through a simulated Gulf Stream modeled current field. The same challenge was posted to an automatic path planning system currently in development. The results of our study demonstrate that the automatic path planning system performed on par with experienced pilots. Furthermore, the testbed environment revealed a problem with the existing ocean current correction system used by the commercially available glider. Using our simulation testbed, we were able to develop and test an alternative heading algorithm.