Smart instrumented training ranges: bringing automated system solutions to support critical domain needs

Abstract

The training objective for urban warfare includes acquisition and perfection of a set of diverse skills in support of kinetic and non-kinetic operations. The US Marines (USMC) employ long-duration acted scenarios with verbal training feedback provided sporadically throughout the training session and at the end in a form of an after-action review (AAR). The inherent characteristic of training ranges for urban warfare is that they are the environments with a high level of physical occlusion, which causes many performances not to be seen by a group of instructors who oversee the training. We describe BASE-IT (Behavioral Analysis and Synthesis for Intelligent Training), a system in development that aims to automate capture of training data and their analysis, performance evaluation, and AAR report generation. The goal of this effort is to greatly increase the amount of observed behavior and improve the quality of the AAR. The system observes training with stationary cameras and personal tracking devices. It then analyzes movement and body postures, measures individual and squad-level performance, and compares it to standards and levels of performance expected in given situations. An interactive visualization component delivers live views augmented with real-time analytics and alerts; it also generates a personalized AAR review in a three-dimensional virtual or mixed reality environment, indexed by automatically extracted salient events and accompanied by summary statistics of unit performance. The approaches presented in the system have the potential to radically change the analysis and performance assessment on physical training ranges and ultimately this type of training itself.

Collective Simulation-Based Training in the U.S. Army: User Interface Fidelity, Costs, and Training Effectiveness

“…In addition, systems have been developed to capture objective performance data in networked SBT and process them rapidly to generate output (e.g., graphs, tables, still or animated views of trainees' actions). These data can be used to provide feedback to trainees in real time (e.g., at critical decision points during training events to scaffold training, i.e., to dynamically adapt training events to trainees' actions), in AARs, or to analyze performance after training events (e.g., Brown et al, 1997;Chen et al, 2007;Gately et al 2005;Sadagic et al, 2013). Some systems not only provide feedback about trainee performance but can generate "whatif" scenarios or demonstrate the causal relationships between trainees' decisions and outcomes (e.g., Chen et al, 2007;Sadagic et al, 2013).…”

Towards automated assessment of team performance by mimicking expert observers’ ratings

“…Automated performance assessment has also been proposed at instrumented training sites, e.g., using simple heuristics that reward predefined tactical behaviors, risk aversion, dispersion, mobility to assess the performance of military tactical maneuvers (Sadagic et al 2013). As useful and robust as that type of evaluation system is, there is no denying that observer-based assessments are more flexible and better prepared to handle indicators that fall outside the scope of the automated system's logging capabilities, e.g., as a consequence of improvisation or unanticipated decisions.…”

“…The classical after-action review (AAR) methodology is an example approach to facilitate such positive knowledge creation (Rankin et al 1995; Morrison and Meliza 1999; U.S. Army Combined Arms Center 2011). It is worth noting that automation has a role also in AAR:s, with researchers striving to integrate an automatically generated baseline/ground truth to focus discussions around (Frank et al 2008;Sadagic et al 2013).…”

“…Such definitions have the benefit of being tangible and quantifiable and are common e.g., in production industry (Molleman and Slomp 1999) and sports (Papps et al 2011;Scoppa 2015). Similar utilitarian perspectives have been used in the military domain to define team performance when assessing training of certain mission critical or highrisk tasks, e.g., military operations in urban terrain (Sadagic et al 2013), aviation (Deaton et al 2007), and cyber security training (Abbott et al 2015;Granåsen and Andersson 2016). Automated measurement of performance in such settings is both commonplace and sound, with the caveat that for such automation to be feasible, and the training sites need to be prepared with infrastructure and technology to collect data and feed it to assessment algorithms.…”

Visual Analysis and Filtering to Augment Cognition

“…BASE-IT, short for Behavioral Analysis and Synthesis for Intelligent Training [1], was developed at the MOVES Institute at the Naval Postgraduate School, the University of North Carolina, Chapel Hill, and the Sarnoff Corp. (now part of SRI). The goals of BASE-IT are to address some of these difficulties and:…”