Description:
OBJECTIVE: Develop hardware and/or software tools to accurately determine full body segment positions and orientations of a person performing various activities in natural indoor and outdoor environments. DESCRIPTION: Full-body human motion capture has a variety of important applications within the Air Force and Department of Defense, as well as in numerous commercial industries, such as athletics, health care, and entertainment. AFRL applications include the creation of biofidelic avatar-based training scenarios and the collection of"ground-truth"data for research on human surveillance and tracking methods. Natural settings, including varying terrain, backgrounds, and clothing, are important to AFRL applications, where multiple sensor modalities are used coincidently (i.e.,"sensor fusion"). For example, motion capture can serve as the ground truth for synchronized radar and video collections, where the background and clothing worn are critical to replicating in-field video feeds and the outdoor terrain is critical to replicating in-field radar returns. Current full-body motion capture technology has limitations that inhibit its use in natural or real-world settings. The current gold standard in accuracy is optical motion capture, which relies on line of sight between multiple cameras with light emitting strobes and retro-reflective markers placed on the subject. Optical systems are, however, cumbersome to move and cannot be used with typical attire. Other motion capture technologies exist, each with their own limitations. Electromagnetic sensors provide accurate orientation and position, but are greatly limited by the range of the generated magnetic field. Inertial measure units (IMUs) increase portability, but are limited to orientation measurements only. Markerless motion capture methods focus on fitting a model to a silhouette extracted from 2-D video but are often inaccurate for precise motion analysis. Additional information, such as from a depth sensor (e.g., Microsoft Kinect) can be used to provide some 3-D information. While advances in these areas have shown promise as a replacement to optical motion capture, to date commercial products that provide sufficient accuracy are still unavailable. AFRL is seeking innovative hardware and software tools that will result in the development of a motion capture technology that is: 1) mobile (can be relatively easily moved to various locations), 2) compatible with a variety of clothing, and 3) not restrictive of natural motion (e.g., untethered/wireless). In particular, we are interested in tools that fuse different hardware modalities. For example, an IMU-based motion capture system might be augmented with markerless motion capture techniques or a local positioning system to create a single motion capture system capable of accurate orientation and position tracking. PHASE I: Develop an initial concept hardware and software design in order to accurately determine human segment positions and orientations under clothing in a natural environment. Demonstrate the ability to design and implement the technology through proof of concept. PHASE II: Develop and demonstrate a fully functional prototype of the hardware/software system. Integrate all hardware so that it can be controlled from a single software interface. Validate the system"s accuracy through laboratory experiments. PHASE III DUAL USE APPLICATIONS: The technology will allow the military to collect ground-truth human motion data in realistic operational environments to assist in human threat detection. The technology will provide researchers and doctors with a motion capture system that can record motion in realistic settings.
