Willow Garage History
Scott Hassan founded Willow Garage in late 2006 to accelerate the development of non-military robotics and advance open source robotics software. Named as an homage to the garages that spawned the computer industry, he envisioned an innovative research lab infused with the engineering expertise of a product development company. The goal was simple: to push the frontiers of robotics, both scientifically and commercially. Scott put together a funding package with the unique goals of impact first, return on capital second, with the strong belief that success in the first goal (impact) would provide plenty of opportunities to excel at the second (return on capital to investors).
Early Projects
Mirroring the field of robotics, Willow Garage was born out of the confluence of academic and industrial research.
In 1997, Scott founded eGroups and personally witnessed the power of communities as his online community site evolved into Yahoo! Groups. He also saw similar power as a key software developer for Google, Alexa, and the Stanford Digital Library.
When he founded Willow Garage in 2006, Scott knew that open source would be key to the company's success. Open source lets research build on research and would provide more value to those looking for Willow Garage to contribute. Most importantly, open source would enable Willow Garage to participate in a larger robotics community. As Willow Garage looked to build new research platforms for communities to grow around, it would need individuals from commercial and academic backgrounds who would were already successful in building robots and communities.
Scott first tapped into the burgeoning DARPA Urban Challenge community, where he found a small team working to build an autonomous vehicle to navigate an urban environment. Together, they built Willow Garage's first garage: a shipping container that housed a Ford Escape Hybrid, which they modified to add drive-by-wire, laser, cameras, computers and an inertial measurement unit.
In early 2007, Willow Garage started a second project to develop an autonomous boat. The goals for the boat were lofty: to travel around the world and to achieve a full year of autonomous operation at sea. The design requirements were equally challenging, requiring a hull sleek enough to move through the water using minimal power, but with enough surface area to support a solar-panel array. With two projects under way, Scott looked for a CEO to guide the growing company.
Building a Team
Steve Cousins had a background that spanned both academic and industrial research: Xerox PARC, IBM Research, Washington University, Interval Research and Stanford. In May of 2007, with Willow Garage still just six employees, Scott looked to Steve's experience managing research to round out Willow Garage's vision. Steve's first mission as CEO of Willow Garage was to find the people to grow it from a fledgling company to a world-class research and development organization. Luckily, he wouldn't have to look far to find researchers experienced in contributing to the robotics community.
Gary Bradski was a consulting professor at Stanford, where he was part of its DARPA Grand Challenge team and also helped kick-start the Stanford Artificial Intelligence Robot (STAIR) project. While at Intel in 1999, Gary started the Open Computer Vision (OpenCV) library and was instrumental in getting it released as open source. OpenCV has over a million downloads and proved instrumental in a defining milestone for robotics: it powered the vision system of Stanford's Stanley car, the first autonomous vehicle to successfully complete the DARPA Grand Challenge.
Keenan Wyrobek and Eric Berger crossed paths with Gary on the STAIR project. In April 2005, as graduate students at Stanford University, Keenan and Eric started pitching the idea of a hardware and software development platform for personal robotic applications. Their vision was to change the model of each lab making their own hardware and software, to allow researchers to share code and build on each other's results. By December 2005, they had raised enough money to build their proof of concept and spent the next year making it a reality. Named PersonalRobotOne (PR1), it integrated capable safe arms, sensors, power system and computation into a robust development platform. With their prototype complete and in use by the STAIR project, they sought funding to turn their prototype into a shared mobile manipulation platform.
Meanwhile, Kurt Konolige was busy down the street at SRI. Kurt has a long career in robotics that has straddled academic and commercial research as a senior researcher at SRI, entrepreneur at Videre Design and consulting professor at Stanford. He invented the Pioneer class of robots to use in his Mobile Robotics Lab course at Stanford, and currently has designed a newer low-cost platform called ERRATIC. In the late 90's he developed some of the first ladar SLAM algorithms, and went on to work with visual odometry and stereo algorithms in DARPA's Learning Applied to Ground Robotics (LAGR) project.
Working with Kurt on the LAGR project was Brian Gerkey. Even as a talented researcher in multi-robot coordination, Brian is best known for his work founding the Player Project with Richard Vaughan, Andrew Howard and Kasper Stoy. While doing their multi-robot systems research at USC in 1999, they were frustrated with the closed-source software on their robots and built a new robot software platform. Most importantly, they released it as open source. Even though they wrote it for the robots in their lab, they soon found contributions pouring in from the broader robotics community to add support for different sensors, platforms, and algorithms. With the support of many, the Player Project became the most successful open source robotics platform and is widely used in both academic and industrial research labs.
In mid 2007, Willow Garage was still small and working hard on the autonomous vehicle and boat, but the pieces were starting to fall into place for its future growth. Scott and Steve were building a team at Willow Garage. Gary had successfully launched an open source project from a large commercial company and was looking for a new company to give it a home. And there was Keenan and Eric, who were looking to build a new platform, as well Brian and Kurt, both already successful in building platforms.
New Platforms
Scott and Steve met with Eric, Keenan and Andrew Ng of the STAIR project and found their goals well aligned. Willow Garage could provide the resources for turning the PR1 from a single prototype into dozens of robots in universities across the world. Andrew and the STAIR project would be able to use that robot to perform cutting-edge research. Thus began the first of Willow Garage's many academic collaborations with Willow Garage funding research in the STAIR project.
At the end of 2007 Eric and Keenan joined Willow Garage to lead the Personal Robotics Program and work began on PR2 (the hardware) and ROS (the software).
Gary and Kurt also joined Willow Garage to lead a new Robot Perception research area and Brian took the lead of the Robot Systems effort. With so many new people on board, it was time for Willow Garage to refocus. The DARPA Urban Challenge had come to a successful close and, even with new employees, Willow Garage found itself stretched thin. The autonomous vehicle and boat projects were suspended and the boat was donated to the University of California at Santa Cruz. Willow Garage's new focus had become clear: deliver hardware and software platforms to the research community to help spur autonomous robotics. The PR2 was an ideal mobile manipulation platform for personal robotics and Willow Garage researchers had already been successful in building open source robotics libraries. Of course, there was much to be done.
The PR2 would have to be designed for manufacture on a larger scale. It would need a software platform to develop and integrate open source robotics libraries. And it would need leading-edge perception capabilities to enable compelling research applications in mobile robotic manipulation.
Keenan and Eric embarked on the new version of their robot, redesigning nearly every component: a new head with pan-tilt stage for mounting sensors; human-scale, 7-DOF, removable arms with a modular gripper attachment; and a new base with a quasi-holonomic caster arrangement.
On the software front, Willow Garage decided to build a new, open-source robotics platform with the goals of promoting greater software reuse and simplifying the task of building complex robot systems. It was named ROS to stand for both Robot Operating System and Robot Open Source.
Even with a strong team, Willow Garage knew that it needed to rely on the community to help build ROS and the software on top of it. Brian teamed up with Willow Garage Engineer Ken Conley and Stanford researcher Morgan Quigley, who built an early version of ROS for STAIR, to lead ROS development. They put all of the software on SourceForge as a first step toward building a community. Willow Garage also took over maintenance of OpenCV and built a team around Gary and Kurt to develop the perception capabilities that will be necessary for meeting the challenges to come.
Now
Willow Garage is still young but its mission is well underway. Keenan and Eric's vision of a common mobile manipulation platform is materializing with three alpha PR2 robots already in testing and ROS development progressing. Scott and Steve are continuing to build the team they envisioned for leading-edge robotics R&D, which is now now about 35 employees. ROS has not reached its first official release, yet, but is already in use in at least four universities. It's an exciting time.
