The exoskeleton is not entirely passive. A small amount of energy is required to control the dampers' variability. (The dampers contain a fluid with tiny magnetic particles. When electricity is applied to the fluid, these particles change its viscosity.) But it is very efficient compared with other such systems. "Our exoskeleton only consumes two watts of electrical power during walking," says Herr. This is nothing compared with the 3,000 watts consumed by a motorized exoskeleton.

But there is a catch. Tests of the exoskeleton revealed that although it lightens the load for the user, that person consumes 10 percent more oxygen than if he or she had simply carried the load without mechanical assistance. This higher metabolic rate is attributed to the fact that the device interferes with the natural gait of the walker. "Walking with the exoskeleton takes more energy than walking without," says Michael Goldfarb, director of the Center for Intelligent Mechatronics at Vanderbilt University, in Nashville, TN.

Even so, it is a good effort, says Goldfarb. "I'm not aware of any exoskeleton--active or passive--that has been shown to effectively decrease metabolic energy expenditure," he says. And even if more energy is burned, the exoskeleton still reduces the stress on the wearer's back and legs.

The MIT group believes that by carefully selecting and angling the springs, it can reduce the amount of energy that a person needs to walk with the exoskeleton.

It would probably take about two years to commercialize this technology, says Herr. "But we have no plans at this time to move forward with commercialization," he says.

Goldfarb still believes that there are hurdles to overcome. There are great advantages to using variable dampers and springs, not least that they are much lighter and less power hungry than motors and actuators, he says. But a device that requires less effort and is capable of covering a broad range of terrains, such as uneven surfaces and stairs, must have not just variable dampers but also springs of variable stiffness. This is a taller order, Goldfarb says.