| Autor: |
Sadagic, Amela, Welch, Greg, Basu, Chumki, Darken, Chris, Kumar, Rakesh, Fuchs, Henry, Cheng, Hui, Frahm, Jan-Michael, Kolsch, Mathias, Rowe, Neil, Towles, Herman, Wachs, Juan, Lastra, Anselmo |
| Přispěvatelé: |
Naval Postgraduate School (U.S.), Computer Science (CS) |
| Rok vydání: |
2009 |
| Popis: |
Military training conducted on physical ranges that match a unit’s future operational environment provides an invaluable experience. Today, to conduct a training exercise while ensuring a unit’s performance is closely observed, evaluated, and reported on in an After Action Review, the unit requires a number of instructors to accompany the different elements. Training organized on ranges for urban warfighting brings an additional level of complexity—the high level of occlusion typical for these environments multiplies the number of evaluators needed. While the units have great need for such training opportunities, they may not have the necessary human resources to conduct them successfully. In this paper we report on our US Navy/ONR-sponsored project aimed at a new generation of instrumented ranges, and the early results we have achieved. We suggest a radically different concept: instead of recording multiple video streams that need to be reviewed and evaluated by a number of instructors, our system will focus on capturing dynamic individual warfighter pose data and performing automated performance evaluation. We will use an in situ network of automatically-controlled pan-tilt-zoom video cameras and personal position and orientation sensing devices. Our system will record video, reconstruct dynamic 3D individual poses, analyze, recognize events, evaluate performances, generate reports, provide real-time free exploration of recorded data, and even allow the user to generate ‘what-if’ scenarios that were never recorded. The most direct benefit for an individual unit will be the ability to conduct training with fewer human resources, while having a more quantitative account of their performance (dispersion across the terrain, ‘weapon flagging’ incidents, number of patrols conducted). The instructors will have immediate feedback on some elements of the unit’s performance. Having data sets for multiple units will enable historical trend analysis, thus providing new insights and benefits for the entire service. Office of Naval Research |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
