A modified version of Doom II is helping US troops test their reactions in simulated combat, though once platoon members are hit by artillery shells, there is no computer-generated medivac team to tend to their wounds.
Sandia National Laboratories wants to come to the rescue. But the research lab's approach should not be confused with a videogame.
The Virtual Reality Intelligent Simulation (VRIS) is a federally funded project that researcher Sharon Stansfield hopes will show that VR is more than a novelty best used in entertainment - that it is a practical, hands-on training tool.
"We need a way to train small teams of people under stressful situations to help them learn how to react," said Stansfield.
Sandia is developing two separate projects under VRIS that will eventually share technologies. The first, VR-MediSim, immerses medic trainees into a virtual battlefield where they must contend with a confrontation and then approach a simulated casualty, and diagnose and treat the virtual patient's wounds. To do this, a trainee moves between two different virtual worlds, governed by artificial-intelligence technology.
In the first world, the trainee is approaching a combat situation - for example, a military team charged with clearing an area is confronted by a sniper, who wounds one of the soldiers.
While the other soldiers work to neutralize the enemy fire, the medic tends to the casualty. The medic trainee then enters the VR-MediSim world to examine the casualty from different perspectives and use virtual medical instruments.
A trainee receives feedback by interacting with the virtual casualty, issuing spoken requests for information or using the virtual medical instruments.
A second Sandia project, VRaptor, is designed to train law enforcement officers in hostage situations. These exercises are so brief, Stansfield said, that artificial intelligence is not needed.
"There are only so many things that can happen in these instances. A hostage can have hands on head, be sitting or standing or falling to the floor dying, so we just need the software to move quickly," she said, noting that these actions are programmed into the simulation.
The only decision for the VRaptor trainees is to shoot or not shoot, and the actions taken will depend on what happens seconds after they enter a room. For example, if trainees break down a door, dust that may result will show up in the simulation and momentarily cloud their vision. Feedback is delivered silent-movie style; if a trainee takes a bullet, a "You've Been Shot!" message flashes.
In both projects, participants don the familiar VR headgear and gloves. In the case of VRaptor, trainees also wear magnetic sensors that enable the simulation to react to their up/down and sideways motion, as well as to the motion of raising a gun.
Neither VRaptor nor VR-MediSim are preprogrammed to the point where a trainee can work through all situations and "conquer" the simulation in the same way a player of Doom II can win the game. The situations are fluid and dependent upon the input from a trainer, said Stansfield.
VRIS is but one of a series of projects under Sandia's Laboratory-Directed Research and Development program. LDRD was created by Congress in 1982 to allow the national laboratories to use 6 percent of their operating budget to fund important, high-risk research. Between 1985 and 1995, Sandia has spent US$298 million on such projects. In two years of work on VRIS, the laboratory has spent $816,000.
Now in its third year, VRIS is moving beyond the proof-of-concept stage and on to practical applications. Stansfield expects VR-MediSim, which has also received Defense Advanced Research Projects Agency funding, to find its way to battlefield testing in another year. VRaptor, which is being developed in consultation with the Federal Bureau of Investigation, is probably about two years out from field use.
