Intelligent Coaching Systems in Higher-Order Applications: Lessons from Automated Content Creation Bottlenecks


Greuel, C., Murray, J., Ziker, C., Yarnall, L., & Kernbaum, A. (2016). Intelligent coaching systems in higher-order applications: lessons from automated content creation bottlenecks. In Proc. of 25th International Academic Conference. International Institute of Social and Economic Sciences, Paris, France.


Intelligent virtual environments hold promise for improving learner-directed instruction in context. These systems trace the progress of learners performing tasks and can insert immediate coaching to focus learner attention, link knowledge to activity, and accelerate the shift from abstract to concrete learning. Such technology has been used to improve self-directed learning of hands-on procedures, but also shows promise for higher-order applied fields, such as engineering.

To realize this vision, research must address the formidable bottlenecks around content creation and build understanding of the types of reusable content libraries relevant to the subject domains. This presentation describes two projects for interactive training that developed prototypes for automated content creation. A third project is presented that illustrates a suite of learning object libraries to support engineering instruction.

The first project, SAVE, uses a 3D browser-based simulation environment not only for hands-on training in equipment maintenance, but also for automating the generation of instructional exercise solutions. SAVE allows a subject matter expert to use the interactive simulation for modeling the correct steps of a procedure, thus providing a rapid way to extract their knowledge. The system collects a trace of the expert’s activity, which becomes the reference against which learner activity is compared in automated assessment.

The second project, AR Mentor, delivers augmented reality overlays in head-mounted displays worn by student mechanics while learning to maintain terrestrial vehicles. An automated speech system interacts with the students as they perform equipment adjustments and troubleshoot system faults. To deliver audible step-by-step guidance, a prototype text-to-speech translator was developed to convert steps as written in the technical manual into the voice of a virtual coach.

The third project, SiMPLE, developed a library of engineering computation objects to allow learners to construct electromechanical simulations, and provides an intelligent coaching system to allow novice engineers to iteratively refine their design specifications. When a working simulation is achieved, the system is linked to a 3D printer for physical prototype production.

The first two projects demonstrate methods of using virtual intelligent technologies to accelerate training content production in hands-on domains: expert model tracing and technical manual translation. The third project provides the tools needed to support engineering instruction: object libraries with embedded computations, as well as scripts for design coaching, design testing, and physical prototyping.

Together, these projects illustrate the wide range of available, reusable libraries and the extensive opportunities for automating content creation in many socio-technical fields.

Read more from SRI