The smart robot conundrum
It takes a village to make a smart robot. But does it take a dictator to found a village?
That's the question dogging Joe Bosworth this week at the O'Reilly Open Source Convention in Portland, as he pushes his Open Robot Network plan for smart robotics development.
Beneath the exterior of XML and robot web services as the heart of the industry's development, his underlying APIs are, and will remain, proprietary. (That's one of Bosworth's current products to the right. Cute, huh?)
"It’s a question of proceeding as a business," he said. "And the way you do that, no matter who you are, is to have something that’s yours in order to start capitalizing on that. Otherwise you’re giving code to someone else."
These are still Heathkit Days for robotics, with a few single-purpose hits like the Roomba and Aibo, but no unifying user interface to tie it all together, and no killer app for a general purpose machine. Bosworth believes that if people license his navigation system, giving robotics a common platform, that open source extensions and applications will begin to flourish, and the world will be changed, as it was by the Apple II 30 years ago.
"There are a number of components that, when they become fused, including the interfaces, then we have the base," he said. But without a revenue base under that foundation technology, it can't develop. Thus robotics needs a Steve Jobs right now, not a Linus Torvalds.
Since these are Heathkit Days, there are other proposed platforms being offered. The Orocos Project for robotic control has been around for five years in Europe. Metris, a U.S., company, acquired Krypton, one of the major players here, back in April. Most of their work today is in the industrial tools market. The Japan Robot Association has also released an Open Robot Interface (PDF), based on Windows.
But, again, these are early days. So open source developers are going to hear some harsh truth tomorrow from Joe Bosworth.
"My general notion is that developing this kind of mechanism is very difficult as a committee project. It’s like designing something very complex and substantial as a committee. I’m not sure you design as a committee. I think you implement that way. You need a strong, single solution that works, which can be demonstrated, which people can add to."
How well will that medicine go down? Joe Bosworth could be on his way toward becoming the real equivalent of the fictional Laurence Robertson. Or he could become another Ed Roberts. Who knows which? That's what makes reality so much fun.