Robot hardware is set to get the open source treatment, with the launch of Robotic Open Platform (ROP), a detailed set of specifications designed to accelerate robot development and reduce costs by providing researchers with a library of hardware designs, electrical schematics, and CAD drawings that they can contribute to and modify.
Launched by a team at Eindhoven University of Technology (TU/e) in The Netherlands, ROP was created to help robot engineers overcome some of the problems commonly associated with independent robotics labs research—problems that range from costly and time-consuming development processes to incompatibilities between different robotic systems.
If successful, ROP could reduce development costs, speed time to market, encourage efforts to standardize software APIs and hardware designs across the robotics industry, and even lead to the creation of universal standards in robotics, similar to those already used in the PC industry, says Heico Sandee, program manager for robotics, Faculty of Mechanical Engineering, at TU/e.
“One of the things we need to do is make robots less expensive. Many companies are focused on one-of-a-kind robots, but you need to sell large quantities of robots to make them cheap. But the large volume of sales isn’t there because the robots are all too expensive. So, it’s sort of a circle that we’re in, and one that we somehow need to get rid of,” Sandee says.
Strength in numbers
The team at TU/e was inspired to create ROP by the success of ROS (Robot Operating System), the open source robotics software platform. ROS, which continues to gain industry acceptance, is effectively a meta-operating system for robots. It provides a range of services, including hardware abstraction, low-level device control, implementation of commonly used functionality, message passing between processes, and package management.
“It’s the idea that with a lot of people working together, you can achieve a lot more, especially on the difficult stuff. In the same way, we want to create a big robotics community where people can share their knowledge, their ideas, and their designs about everything that they have on their robots, in order to work together to achieve better and cheaper robots,” says Sandee. TU/e has already added the complete blueprints of its AMIGO humanoid care robot and will soon add the designs for its soccer-playing TURTLE bot.
If enough people contribute designs for specific robot hardware components (such as arms and moving bases, for example) to ROP, says Sandee, eventually a cheap, open source robot could be manufactured at a greatly reduced cost. That’s a powerful argument. Today, a fully tricked out AMIGO robot would cost around 350,000 euros. The team at TU/e wants to make it possible for companies to build an AMIGO for around 10,000 euros within five years.
There’s precedent for the researchers’ goal. One robot that has benefited from embracing an open source hardware model is the e-puck, a 7-cm high, differential wheeled robot developed by a team at EPFL in Switzerland. Companies from Switzerland, North America, and Japan now manufacture the robot (or components of it), demonstrating that the open source model can create commercial opportunities for developers.
Little wonder that several robotics companies have expressed an interest in contributing designs as well. Willow Garage Inc., Menlo Park, Calif., has added its TurtleBot to the mix. Other contenders include Germany-based KUKA, with its youBOT (list price is 23,990 euros) and DEMCON, with a set of robotic arms. Sandee hopes to have six to 10 full robot hardware designs available on the site within the next six months.
Robotics companies that share their hardware designs will benefit from the exposure. The ROP community will tweak their designs and speed up product development, says Sandee. “Designers always want to have a better robot. If they put their designs online and the community starts working cooperatively on them, the company can then use the results to create a better robot,” he explains. “Of course, it takes some time within these companies to work out what exactly they can contribute, but we’ve already had some companies giving us very positive feedback.”
ROP fills a major need by providing common hardware platforms for the robotics researchers and developers, says Ken Conley, ROS platform manager at Willow Garage. In 2010, Willow Garage donated 11 PR2 robots to research institutions worldwide as part of its PR2 beta program. The program, due to end next year, has resulted in some remarkable applications, from the P.O.O.P.S.C.O.O.P. to a laundry-folding robot.
ROP will benefit researchers by enabling them to build common software platforms on top of off-the-shelf designs, says Conley. “Researchers can then focus on novel hardware contributions rather than reinventing the wheel. Common robotics hardware can also improve science in robotics research, by facilitating the comparison of techniques,” says Conley.
Robot hardware platforms are often constructed using off-the-shelf mobile bases, sensors, and manipulators. ROP could simplify the integration effort and also encourage standardization in components assembly, says Conley, which would enable ROS developers to focus on fewer shared hardware platforms instead of many custom platforms.
“Much like the IBM PC accelerated the development of the personal computer industry, common robotics hardware lowers development costs [and] enables software developers to focus on shared systems. We’re really excited by the ROP effort and look forward to seeing the ROP and ROS communities grow together,” says Conley.