Wizards of ROS: Willow Garage and the Making of the Robot Operating System
Ten years ago today, an engineer at Silicon Valley robotics lab Willow Garage posted a new code repository on SourceForge. The repository made publicly available to anybody in the global who wanted to get admission to it hosted the codebase for a brand new project Willow became working on ROS. The ROS code repo, installation by using Ken Conley, ROS platform manager at Willow, on November 7, 2007, at 4:07:forty two p.M. PT, changed into the primary time the time period ROS became used as a formal, public designation for Willow’s Robot Operating System challenge.
Choosing a specific date for the tenth anniversary of ROS is a bit complicated due to the fact what we recognize as ROS today is both older and younger than this: The idea of a robotic running machine began at Stanford University, advanced through Willow Garage, and now is living with Open Robotics. It’s a complicated tale that has formed plenty of the robotics industry in recent years. As robotics research makes the difficult transition to corporations and products, the impact of ROS is becoming even extra said.
Over the ultimate numerous weeks, IEEE Spectrum has been speaking with some folks who helped shape ROS, from its origins as part of Stanford’s Personal Robotics Program to Willow Garage and its PR2 Beta Program and beyond. Of course, many different people contributed to ROS, and we weren’t capable of talking to them all. This is our initial effort to put together oral history and tell as much of the story of ROS as we can within the words of the folks who had been there, making it appear.
What follows are excerpts of our interviews with 8 people:
Eric Berger was software lead at the Stanford Personal Robotics Program and director of the Personal Robotics Program at Willow Garage. He is now CTO of Desmos. Keenan Wyrobek becomes mechanical systems lead on the Stanford Personal Robotics Program and director of the Personal Robotics Program at Willow Garage. He is now head of product and engineering at Zipline. Morgan Quigley changed into a PhD pupil at the Stanford AI Lab. He is now chief architect at Open Robotics. Steve Cousins changed into CEO of Willow Garage. He is now CEO of Savioke. Brian Gerkey becomes director of open-source improvement at Willow Garage. He is now CEO of Open Robotics. Melonee Wise became a senior engineer and supervisor of robot development at Willow Garage. She is now CEO of Fetch Robotics.
Tully Foote turned into software program improvement supervisor at Willow Garage. He is now the ROS platform manager at Open Robotics. Ken Conley become ROS platform manager at Willow Garage. He is now an engineer at Aurora Innovation. We grouped their remarks into sections, seeking to prepare matters thematically as well as chronologically, although a few occasions may also overlap or span many months or years:
1. Stanford, STAIR, and PR1
2. Scott Hassan and Willow Garage
3. Willow’s Personal Robots Program
4. ROS and PR2
5. ROS inside the Open
6. Beyond Willow
1. Stanford, STAIR, and PR1
Eric Berger: Before ROS itself changed into a concept, Keenan Wyrobek and I have been operating in Ken Salisbury’s lab at Stanford, going for walks on a mission known as the Personal Robotics Program. We had two matters that we were seeking: Build a hardware platform and build an open-supply software program equipped with the fundamental goal of building a robotics improvement platform. We were grad students, the issues we noticed around us were grad student problems, and we saw grad students in robotics losing an entire lot of time. People who’re suitable at one part of the robotics stack are usually crippled via every other element—your undertaking making plans is good, but you don’t recognize anything about imaginative and prescient; your hardware is the first-rate. However, you don’t recognize anything about software programs. So we got down to make something that didn’t suck in all of these exceptional dimensions. Something that changed into a respectable place to construct on top of.
“We saw grad students in robotics wasting an entire lot of time. Some other part commonly cripples people who are excellent at one part of the robotics stack… So we got down to make something that didn’t suck.”
Keenan Wyrobek: We talked to individuals who did surprising PhDs in robotics, and we saved hearing this identical topic: “Yeah, I spent 4 years hacking together a robot and a gaggle of code, I slightly were given it operating, I made a video, and that was the top of that.” We heard that time and again. At some factor, Eric and I have been similar to, what’s truely important is solving this larger trouble that humans are speaking about. I think both people hate losing their time. Maybe we’re lazy could be another way of putting it; however, the notion of spending four years doing what someone else had performed earlier than… we couldn’t get enthusiastic about it. That’s wherein it started: Could we honestly create something that could allow humans to build on each different’s outcomes in place of continuing this cycle of 90 percent duplicating what someone else has already performed, with a touch bit at the quit of something? Now, if you’re lucky.
PR1 robotic prototype Image: Stanford
The PR-1 prototype, a cell manipulation development platform, becomes constructed with Keenan Wyrobek and Eric Berger as part of the Stanford Personal Robotics Program at Ken Salisbury’s lab.
Eric Berger: We built a hardware prototype, the PR1, and we started running on a software program to try and get it to move. That was the unique thing that had the call ROS on it, although I don’t assume it truely shared any code with the thing that ended up shifting forward. We spent quite a little time considering what it takes to make robots paintings properly and write down how ideally we’d want to make this code. We regarded round at a gaggle of things—we took numerous notions from something that Morgan Quigley has been operating on known as Switchyard.
Morgan Quigley: Switchyard became our first attempt at having the STAIR Project paintings collectively. Photo: STAIR Project/Stanford The STAIR 1 robot was built using Andrew Ng’s group at Stanford as part of the Stanford Artificial Intelligence Robot (STAIR) task. STAIR 1 robot was constructed using Andrew Ng and his group at Stanford. The goal was to produce a viable home and workplace assistant robot. STAIR was an effort that Andrew Ng started, even earlier than I was given to Stanford. It has become within the tradition of these grand college AI and robotics initiatives that pass all the way lower back to Shakey and Flakey. The idea of STAIR became to unify AI initiatives into one unmarried robotic. There were many distinctive additives to the STAIR venture—imaginative and prescient, manipulation, navigation… And anyone wanted the one’s matters to talk to every other at the equal robotic.
One of the demanding situations in those huge organization tasks is that through nature, everyone thinks that their a part of the project is the maximum critical element. So, everybody tends to want their part of the mission to be freely iterating, but anyone else’s stuff needs to live nevertheless and now not be a shifting goal to combine with. The concept of Switchyard, and ultimately ROS in trendy, turned into that you could have those subcomponents which might be changing all of the time. However, you could version them to maintain the relaxation of the device constant, and that view of the machine may be exceptional for each person the usage of it.