- By Thomas R. Temin
- Mar 30, 2004
Thomas R. Temin
Watching those robotic contraptions run off the road and flip over at the DARPA Grand Challenge last month might have been amusing. But the Defense Advanced Research Projects Agency and the teams trying to meet its requirements made significant progress in converging sensors, data networks and machinery.
Like early powered flights, those desert vehicle tests showed the beginnings of what is possible.
Unmanned vehicles have been around for decades. Mostly the technology is found in lift trucks and stock-picking equipment in factories and warehouses. They're guided by sensing magnetic or colored stripes painted on the floor.
Obviously, moving unmanned vehicles outdoors across uncontrolled terrain is infinitely more complex. But it's the best way to see if data from many networked sensors can be harnessed usefully. Future U.S. military doctrine depends on this capability, as does the ability to respond to all sorts of possible homeland security threats that are better checked out by an unmanned device.
What amazed me is not how short a distance some of the vehicles went, but how far.
DARPA is also trying to coax the next generation of network protocols via competition. 'The packet network paradigm probably needs to change,' DARPA's Col. Tim Gibson told GCN's Joab Jackson during the recent Mojave Desert technology bake-off.
Essentially, he said, industry must come up with a scheme to automatically make dynamic quality-of-service allocation to thousands of users and devices. A tall order.
Even the venerable Open Systems Interconnection seven-layer model is open to revision, DARPA officials said. That's because, they believe, it is not entirely suitable to wireless data networks in dynamic environments.
So anyone who thought IP was the final networking scheme'or that tanks carrying soldiers'was the only way will need to revise their thinking.