Two years ago, in his book “Rocketeers,” Michael Belfiore celebrated the pioneers of the budding private space industry. Now he has returned to explore a frontier closer to home. The heroes of his new book, “The Department of Mad Scientists,” work for the Defense Advanced Research Projects Agency, better known as Darpa, a secretive arm of the United States government. And the revolution they’re leading is a merger of humans with machines.

The revolution is happening before our eyes, but we don’t recognize it, because it’s incremental. It starts with driving. Cruise control transfers regulation of your car’s speed to a computer. In some models, you can upgrade to adaptive cruise control, which monitors the surrounding traffic by radar and adjusts your speed accordingly. If you drift out of your lane, an option called lane keeping assistance gently steers you back. For extra safety, you can get extended brake assistance, which monitors traffic ahead of you, alerts you to collision threats and applies as much braking pressure as necessary.

With each delegation of power, we become more comfortable with computers driving our cars. Soon we’ll want more. An insurance analyst tells Belfiore that aging baby boomers will lead the way, enlisting robotic drivers to help them get around. For younger drivers, the problem is multi tasking. Why put down your cellphone when you can let go of the wheel instead? Reading, texting, talking and eating in the car aren’t distractions. Driving is the distraction. Let the car do it.

That’s where Darpa comes in. Belfiore traces the agency’s origins and exploits from the 1957 Sputnik launching (which shocked the United States government into technological action) to the 1969 birth of the original Internet, known as Arpanet, to Total Information Awareness, the controversial 2002 project that was supposed to scan telecommunications data for signs of terrorism. His tone is reverential and at times breathless, but he captures the agency’s essential virtues: boldness, creativity, agility, practicality and speed.

The Army needs vehicles that can move cargo without exposing human operators to bombs or enemy fire. To encourage development of such vehicles, Darpa sponsored a 2007 contest in which cars designed by 35 teams navigated a simulated urban war zone. The cars used systems like those already in consumer vehicles: GPS, lane guidance, calibrated braking. But instead of routing their information and advice through human drivers, the cars simply acted on it.

Belfiore recounts several low-impact crashes caused by the limited ability of current software to understand complex traffic situations. But with each successive contest since Darpa’s first robot-car race, the Grand Challenge, in 2004, performance has improved. In some respects, the robot cars already surpass us. Their reaction speed is better. They can see at night, thanks to laser range- finders. They have no blind spots. And when networked, they can read one another’s intentions.

So maybe we’ll let robots drive our cars. But would you let a robot cut you open? That’s Darpa’s next project. In minimally invasive surgery, doctors insert very thin instruments through keyhole-size incisions. This minimizes pain, blood loss, infection risk and recovery time, but it’s hard. Surgeons have to manipulate their instruments indirectly and watch them on a video monitor. They might as well use a machine. It could execute their commands, give better video feedback and hold the instruments more steadily.

More than 850 hospitals already use such operating machines. Surgeons sit across the room from patients, connected to their instruments by game-style controls and three-dimensional video binoculars. When the machines meet resistance, the surgeons feel it. The goal is to engage the doctors’ senses as fully as if the mechanical eyes and hands were theirs. In fact, they are theirs. The surgeons’ minds map, orchestrate and experience the machine like an infant taking possession of its own body.

But if sensory feedback can extend a surgeon’s body across a room, why stop there? A new version of the machine adds Ethernet, freeing the doctor to inhabit a mechanical body anywhere with a good cable or wireless connection. By digitizing surgical commands, we’ve already created transitional moments in which maneuvers have been described but not executed. Why not extend this transition, playing out the surgery in virtual reality and then editing out any errors? That’s the next step: surgery with a word processor, so to speak, instead of a typewriter.

READ MORE