[1]This material has been taken from [28]. The full text appears in Appendix A.
The NPSNET Research Group has begun looking at the complete set of issues regarding integrating the human figure into the DIS synthetic environment. The driving problems of our current efforts are the integration of dismounted infantry [24] and medical corpsmen into the DIS virtual environment. Some of the issues being addressed include: the robust representation of humans in the DIS synthetic environment, the multi-resolution networking of human interactions, and the addition of physical, wounding models into the DIS synthetic environment. Recent accomplishments are identified here, while Appendix A presents additional near-term and long-term milestones of the NPSNET-Human project.
One of the first things that one needs in putting a human figure into a VE is the ability to display an accurate, articulated human viewpoint. The NPSNET-Human project has recently completed work on this so that view position/orientation are now computed based on body posture and joint articulations. This means that one's virtual body parts are clearly visible and the user's eyepoint is updated intuitively while the head is tracked with an HMD sensor. This viewpoint improvement allows for realistic viewing at close range, e.g. when a medic looks downward to attend to a victim. It replaces the earlier approximated view which was based solely on posture and assumed only small deviations in head/neck angles.
In early versions of NPSNET, a wounded human was a dead human, with a dead 3D icon displayed for that figure. We have recently begun an effort to provide an intermediate display. We now have networked simple wounding of soldiers in the virtual battlefield. Soldier icons display simple networked wound representations on their body when and where they are hit. In the lower resolution battlefield environment, the wounds adequately indicate the need for medical attention and the location of the injury. Multiple hits cause multiple wounds to be displayed. Currently, only a ballistic type of wound is displayed.
For the medical corpsman scenarios we are developing in NPSNET, we have begun prototyping a networked, autonomous evacuation unit. Companion software has been developed to dispatch a networked, autonomous ambulance to the location of a wounded soldier in NPSNET. Upon reaching the victim, a two man stretcher team disembarks from the rear of the ambulance, walks to one side of the victim and kneels to pick him up. The wounded soldier (for now) is instantaneously transferred to the stretcher and is carried back to the ambulance. The ambulance returns to the aid station from which it was dispatched.
Work underway in the NPSNET-Human project towards the goal of instrumenting the human body is currently focused on real-time arm articulations. An interface has been developed to give a user with Polhemus magnetic sensors attached to his arms the ability to control an icon's arms in real-time. This is a low-cost alternative to the Sarcos ISMS and Sensor Suit. A user can interact with the synthetic environment independent of limited scripted animations. Efficient multicast networking means have been developed to receive/send high frequency data from the user interface tracking real-time upper body motions. The networking of the arm articulations is carried out by a ``high resolution'' parallel-to-DIS network. This parallel network is needed to reduce the amount of articulation data placed onto the net as would be required were standard DIS utilized.
We have started work on investigating the use of realistic wounding models for the human figures in NPSNET. We have begun to identify what models exist and are readily available, and their interface requirements, i.e. parameters regarding weapon type, firing range, projectile's direction of travel, as well as the soldier's posture and his protective equipment.