Extending Persistent Monitoring by Combining Ocean Models and Markov Decision Processes

Ocean processes are complex and have a high variability in both time and space. Thus, ocean scientists must collect data over long time periods to obtain synoptic views and resolve multidimensional spatiotemporal variability. In this paper, we present a methodology for incorporating time-varying currents into a Markov Decision Process for persistent path execution by underwater gliders. The application of an hybrid Gaussian distribution of ocean currents and a modified Markov Decision Process technique enables the incorporation of uncertainty from a deterministic ocean model. The proposed approach achieves improved navigational accuracy, and can extend the distance travelled over the duration of a mission. We present a derivation of our methodology, an outline of the proposed algorithms, and simulation predictions that are validated through experimental field trials.