by DAVID AXE
In May 2010 Willow Garage, a startup robotics firm in California’s Silicon Valley, hosted a graduation for some of the world’s first humanoid robots. Eleven PR2 robots — two-armed ‘bots with wheeled bases, sensors and open-source software operating systems — “danced” with human partners.
The PR2’s debut sparked a minor robot revolution as other research groups acquired the ‘bot, by purchase or donation, and modified its open software. “Within just months of receiving its free PR2, a lab at the University of California, Berkley wrote code that allowed the robot to fold towels and bundle socks,” writes Ryan Calo, a Stanford University researcher.
To Calo, PR2 is a model for robotics development, whereby basic robots can be quickly, cheaply and legally modified for a wide range of tasks. “Innovation within open robotics could move at a dramatically faster pace and lead to an accompanying vibrant market for third party software, components and accessories,” Calo writes in the Maryland Law Review.
An open robotics economy would have huge implications for the global economy — and also for the armies of advanced nations. Open-source computers and software have already spurred the development of military “apps” including battle-tough smartphones. The U.S. military’s efforts to encourage robot development also hinge on the open-source model. The Defense Advanced Research Projects Agency, or Darpa, has hosted several robot races since 2004 that have seen the military, universities and industry sharing software, hardware and design efforts. Willow Garage got its start during a Darpa competition.
Spin-offs so far include the Army’s Convoy Active Safety Technology robot trucks, the Marine Corps’ handheld tablet-style controller for K-MAX robot cargo helicopters and a drone-control software suite being developed for the Navy by the Massachusetts Institute of Technology.
But there’s a downside to open robotics that could significantly slow the pace of advancements, Calo warns. “I will predict that open robotics will confront an additional hurdle: the potential for crippling legal liability, which may lead entrepreneurs and investors to abandon open robots in favor of robots with more limited functionality.”
Because they are designed to interact with the physical world, “robots are in a position to cause physical damage and injury directly,” Calo writes.
An out-of-control robot, perhaps following programming from developers other than the original manufacturer, could expose the manufacturer to expensive liability. “The resulting legal uncertainty could discourage the flow of capital into robotics.”
That’s perhaps what motivated Sony when it tried to control work on its robot “pet” AIBO six years ago, Calo explains:
AIBO began as a closed system that ran only proprietary software — AIBOware — that allowed users to “raise” AIBO over time and teach it certain voice commands. Sony eventually published a programming code (“RCODE”) that permitted users to teach the AIBO new behaviors. Users loved R-CODE and quickly bypassed the controls Sony had in place in order to share AIBO programs with one another online. Many people did so, leading to an entire AIBO subculture.
Sony learned of the practice and was not pleased. The company sent a cease-and-desist letter to the popular AIBO forum AiboHack, asking the Website to take down the traded code as a copyright violation. Sony arguably never recovered from the resulting consumer backlash, and it shut down the AIBO line in 2006.
To encourage continued open robotics development, “the time to think through this problem is now,” Calo advises:
First, we should consider immunizing manufacturers of open robotic platforms from lawsuits arising out of users’ changes to robots, at least temporarily. Second, we should consider whether robot owners can carry insurance against the possibility of accidents.
The goal is to keep open robotics alive and thriving, for the benefit of markets and militaries.