Automating Adaptive Training for Ship Handling
Abstract
Project Summary The proposed 12-month research effort has three objectives: (1) To enable Naval officers to learn ship handling better and more efficiently, we will integrate extensive adaptive tutoring into COVE ITS 1.0, an artificially intelligent tutoring system (ITS) that provides student conning officers with a series of lessons in handling a DDG-51 class ship simulated in the Conning Officer Virtual Environment (COVE) simulator. Implementing adaptivity of tutoring will exploit the already interfaced Automated Assessment Engine (AAE) for ship handling that the National Center for Research on Evaluation, Standards, and Student Testing (CRESST) at UCLA is developing and extending with support from ONR. We will study where adapting feedback to individual learners and automating variation in the problems presented to them enhances learning–measured by rate, retention, and generalization in the schoolhouse at the Surface Warfare Officers School (SWOS) in Newport, RI. This research is expected to lead to an advanced COVE ITS 2.0, which can be incorporated into the Navigation and Ship handling Training and Assessment Tools (NSTAT) 2.0 training system that SWOS is planning as a follow on for NSTAT 1.0. (2) To aid students in their initial experience with each ship handling lesson, and to help them overcome learning plateaus, we will develop a capability for the COVE ITS software (i) to conn a ship as humans do (e.g., sensing the effect of current on a ship’s movement by observing the ship’s drift relative to its heading and its wake–rather than reading out current data from a simulator or calculating the data from a GPS/inertial reference system) and (ii) to explain its decisions and actions in spoken English. COVE ITS 1.0’s ship handling knowledge representation will be reorganized and expanded so the software utilizes information about ship handling and a ship’s environment within the limits of human performance constraints. We will expand on COVE ITS’s current capability of driving a simulated ship through a channel transit to enable the system to moor a ship at a pier and to conduct underway replenishment. We will evaluate the value of such demonstrations for students of ship handling at multiple stages of the learning process. This automated capability to conn a ship while explaining the process is expected to be incorporated into NSTAT 2.0. (3) To enable assessment of conning officers’ compliance with Rule of the Road (RoR) during evolutions such as harbor transit, we will couple the RoR trainer being developed at the University of Nevada, Reno (UNR) with COVE ITS. We will study the merits of tutoring students on RoR during lessons in ship handling. The connected COVE ITS and RoR trainer systems are expected to be integrated into NSTAT. This research further provides the potential to develop the coupled systems further into a low cost, portable, medium fidelity ship simulator with intelligent tutoring capability, which could make tutored ship handling practice widely available in the fleet
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 12, 2016
- Source ID
- N000141512147
Entities
People
- Stanley Peters
Organizations
- Office of Naval Research
- Stanford University
- United States Navy