Robot as Servant

  |    
技術の歴史

Command Line Heroes • • Robot as Servant | Command Line Heroes

Robot as Servant | Command Line Heroes

About the episode

The 1980s promised robotic servants were in reach. They’d clean up our houses. Bring us drinks. Usher in an era of leisure. We didn’t get robot maids. But if we look around, we’ll find an army of robotic servants already automating away domestic drudgery.

Richard Rowland recounts the extent to which Androbot over-promised on its ability to build a robot servant. 40 years later, we still don’t have robot butlers. Monroe Kennedy III walks us through the complexities of seemingly simple tasks. To make things more difficult, each attempt to build a robot had to build the hardware AND write the code from scratch. Keenan Wyrobek explains that’s why he helped write and share the Robot Operating System(ROS). Leila Takayama describes how beneficial ROS was to the field of robotics. And Terry Fong shares how NASA is using ROS to build the robots that explore our solar system.

Command Line Heroes Team Red Hat original show

購読する

Subscribe here:

Listen on Apple Podcasts Listen on Spotify Subscribe via RSS Feed

トランスクリプト

Where are my slippers? Right here, sir. Thank you, dear. You're welcome, sir. Hey, give me a beer, would you? My pleasure. Did you ever watch The Jetsons? A very mid-century view of the future. And each episode, at some point, Rosie the robot would wheel in and help the family out with some domestic chore. She might make a wisecrack. But basically, Rosie was an obliging artificial intelligence whose whole purpose in life was to wipe down windows, vacuum the carpet, and fetch your slippers. That was the fantasy, a pretty sexist fantasy. And it was our idea of what a robot could be. A domestic servant, a maid that was made of metal. I'm Saron Yitbarek and this is season eight of Command Line Heroes, an original podcast from Red Hat. This season, we're tackling one question from every angle. What is a robot? Since the word was coined almost exactly a century ago, the idea of robots has fascinated us and sometimes terrified us. Science fiction gave us robots that help and robots that rebel. In our novels and movies, robots embody every hope and fear we've got about technology. People rode monorails to world fairs where the 21st century was laid out as one big bonanza of convenience and high-tech magic. But now that our robot future is coming to life, what are they really like? What have we actually made? What might they become? While making this season, we discovered a robot reality that's pretty removed from the robots we imagined. And yet, all those misguided ideas about what a robot should be actually shaped what robots are. And no robot fiction shaped things more than the story we told ourselves about robotic servants. We thought they'd become our butlers and maids, machines like Rosie on The Jetsons. That idea reached a fever pitch in the mid '80s. Maybe because our tech had evolved just enough so we could produce a calculator on wheels. We got a little excited, for the first time we could build the image of the robots we've been imagining. That was enough to make us think, "Any day now robots will be part of ordinary domestic life." In 1983, it was very believable that we would have personal robots. Rick Rowland lives in Las Vegas, where he manages an RV dealership. But Roland has a side hustle, he builds robots, and he's become a kind of robot historian too. He's lost count, but there's upwards of 200 specimens in his collection. Robots from an earlier time, when each new model seemed to herald a robot revolution. His robot collection, full of dusty wheels and plastic faces, reminds us that back in the '80s expectations were high. The future was fantastic. In my lifetime I had gone from no cell phones, no computers, no nothing. Then when I started college, they're using the old-fashioned mainframe where you don't even have a screen, it's just a paper coming out of the machine. Just in those few years, we went to buying a home computer over at Kmart or Gemco and they did amazing things. So a robot should be able to just follow along since a robot is basically a computer on wheels. Every year, Rowland would visit the big Computer Electronic Show (CES) that came to Vegas. In the '80s, conventions like that were stoking the public's desire for robots. You could cruise down the aisles and find all these hopeful, but usually dead-end, inventions. Every new robot company was promising a consumer product that could roll around your living room, serving canapes and dusting the chandelier. Unfortunately, the marketing department worked a lot harder at it than the production department. The company with some of the most enticing promises was called Androbot. It was founded by Nolan Bushnell. You might recognize that name. He's the same guy who founded Atari and his robots were more like video games than anything else. Bushnell produced a machine called BOB, which stood for Brains on Board. There was another model called Topo. These were boxy R2-D2 style robots whose real purpose was to excite the crowds. Our robot expert doesn't let this history die away. He saved the pamphlets he collected at those old CES shows. We got him to read us from an Androbot pamphlet from 1983. I grabbed a bunch of them just in case I had to get a memory jog. "The year is one AB," and of course the AB was for Androbot. "Open a new era in technology history. A dream has engaged man's imagination for centuries, a mechanical person to help with difficult and boring tasks. Once again, the dream is reality. Androidbot Corporation is pleased to introduce you to the first personal robots. Robots that can carry messages and things. Robots that can turn on lights and watch over your home. Robots that can be taught new behaviors. Affordable robots that will grow with you and your family." It was all pretty aspirational because here's the reality. The tech just wasn't ready. Androbot would be out of business by the end of the '80s. Rowland had one of their models, the BOB, and he could tell us from experience that it was basically a walking disappointment. BOB had voice recognition but only 100 or so words and they were mostly commands like, "BOB, move forward," or "BOB, turn left." Mobility was its only real ability. It could move around and avoid obstacles with a sonar range finder. The infrared sensors were set so low that it would constantly run into chair legs. BOB really couldn't serve you a drink or pick up a sock unless you built a ramp up to whatever it was that you wanted it to interact with. So yeah, BOB was kind of a bust. Androbot and dozens of other companies had promised us robotic servants, but the fundamental problem was physics. Making a machine that can navigate your living room and then also hand you a beer? Way more complicated than anybody realized. Making a robot is one of the most interdisciplinary challenges you could ever take on. Monroe Kennedy III is an assistant professor of mechanical engineering at Stanford, and he specializes in robotics. When I think about the complexity of robotics, I like to think about it as a system. You have to understand the mechanical structure, the hardware. You have to understand the electrical component, so putting sensors and actuators onto those mechanical systems. You need to understand the algorithms and the computational units that are going to go and process the sensor information and then command the actuators to do the right thing. And then, once you design that system, you have to understand how the robot is going to be able to interact with real environments and real humans. Each piece of that puzzle is a specialty unto itself. In the '80s, startups like Androbot were basically trying to be experts in every field all at once. Think about it from a software perspective. Back then, you couldn't open source your problems. There was no GitHub. There was no creative commons. If you were designing a new robot, you started from square one. You literally had to design everything, the embedded system, the computational architecture, how the information flowed from the sensors, to the microprocessors, to the actuation systems. There was really no software infrastructure that allowed you to separate, "I'm great at making arms," from, "I'm great at making the walking legs," from, "I'm great at making the vision system." In other words, you had to be great at everything. Plus, consider this, robots are expensive. It might cost you 100 grand to build a robot capable of wiping down a window. Meanwhile, you can hire a human to clean your whole house for 100 bucks. So there's an enormous economic challenge on top of all the technical ones. For decades, that was the situation. Robot development was stymied by the reality that everybody had to start from scratch and robot labor was never going to compete with human labor. Then, something amazing happened. A new way of thinking about robots that would fundamentally change what's possible. In 2006, the personal robotics program got a boost from an unusual source. Scott Hassan was one of the original coders on Google's search engine. He sold his shares for an enormous profit and decided he was going to use his Google money to create the domestic robots that the world had been dreaming about since the 1950s. Hassan started an outfit called Willow Garage. Their plan was a robot called PR2, which stood for Personal Robot 2. The PR2 was designed to work in ordinary home environments. It was this awesome machine that looked like it was built out of an erector set. It could fold laundry, it could recognize objects, it could bring you a beer. All those things that BOB was supposed to do. The PR2 was an 400,000 dollar robot, not exactly flying off the shelves. But for the Willow Garage team, the PR2 wasn't about sales. It was about proving a point. When we set out to build the PR2, we didn't really set out to build a hardware platform. What we were trying to do was figure out what would it take to build a software platform that would enable the applications that people actually want. Keenan Wyrobek was a director at Willow Garage. The team was realizing that what had been missing from robotic development was a unifying software platform that anybody could use. They decided to build that platform and share it with the world. They called it the Robot Operating System, or ROS. We wanted to commoditize the hard parts of robotics, like planning the path that the arm should take, or processing the 3D point cloud data coming off of a laser range finder, so that people could focus on the things that were really new and unique to their application and leverage all the rest of the work that had been done by every robot team. ROS was like a missing piece of the robot puzzle. Suddenly, you didn't have to be an expert in ten different fields to build a robot. You could focus on the one thing you were good at and leave the rest to the community. ROS is middleware. It's a framework that brings together all the different software components that a robot needs. It handles communication between different parts of the system, it provides standard APIs, it includes libraries of commonly used functions. Most importantly, it's open source. What ROS enabled was for robotics researchers around the world to, instead of spending months or years to build the very, very basic building blocks of robotics, they could leverage work that the larger community had already done and instead focus their energy and their time on the novel things that they were trying to achieve. Leila Takayama is an associate professor in the Baskin School of Engineering at UC Santa Cruz. She studies human-robot interaction and was part of the team at Willow Garage that developed ROS. For her, the creation of ROS represents a fundamental shift in how robots are built. I think the key insight that the founders of ROS had was that we needed to have people who are good at building arms focus on just building arms. And the people who are good at building vision systems focus on building vision systems. And the people who are good at building navigation systems focus on navigation systems. And then we needed a way to integrate all of these different components together. Before ROS, every robotics team was essentially reinventing the wheel. They'd spend months or years building basic functionality that dozens of other teams had already built. It was incredibly inefficient. If you think about the history of personal computing, in the very early days, people would hand-wire their computers, and they would hand-wire every connection, and they would hand-build the power supplies, and they would hand-write the operating system. But at some point, we commoditized those pieces so that you could buy a computer and you could buy an operating system and then you could focus on writing the application that people actually wanted. ROS did for robotics what operating systems did for computing. It provided a standard platform that everyone could build on. And because it was open source, improvements made by one team could benefit everyone. One of the most beautiful things about ROS is that it created this culture of sharing and collaboration in robotics. Before ROS, robotics labs were very secretive. Everyone was trying to solve the same problems independently, and nobody was sharing their solutions. ROS changed that. The impact was immediate and dramatic. Robotics labs around the world started adopting ROS, and the pace of innovation accelerated rapidly. Complex robots that would have taken years to build could now be developed in months. The ecosystem effect that you get when you have a common platform is really, really powerful. You get this positive feedback loop where more people using the platform means more tools and libraries available, which means it's easier for new people to adopt the platform, which means even more tools and libraries. But ROS wasn't just about making robot development easier. It was about enabling entirely new kinds of robots. When you can leverage the work of hundreds of other developers, you can focus on much more ambitious projects. Before ROS, most robots were built for very specific, narrow applications. A robot that could vacuum your floor, or a robot that could weld parts in a factory. But with ROS, you could build robots that were much more general-purpose, much more adaptable. The PR2 was a perfect example of this. It wasn't designed to do just one thing. It was designed to be a platform that could learn to do many different things. And because it ran on ROS, researchers around the world could contribute to its capabilities. We ended up giving away eleven PR2 robots to research institutions around the world, and we said, "Here's this robot, here's the software platform, now go build the applications that you think are important." And the results were amazing. Those research teams used the PR2 to explore everything from elder care to restaurant service to scientific research. Each team built on the work of the others, creating a virtuous cycle of innovation. What was really exciting was seeing how quickly the capabilities of these robots improved. Because everyone was sharing their code, a breakthrough made by one team could be immediately adopted by all the other teams. It was like having the entire robotics community working together on the same project. The success of ROS attracted attention from some very high-profile organizations. In particular, NASA saw the potential of open source robotics for their space exploration missions. This is Terry Fong. I am the chief roboticist at the NASA Ames Research Center in California. NASA has been using robots for decades, from the Mars rovers to the robotic arms on the International Space Station. But traditionally, each mission required building custom software from scratch. In the past, we would develop software for each mission independently. Every time we sent a robot to space, we'd have to solve the same basic problems over again. Navigation, manipulation, perception. It was incredibly inefficient. When NASA discovered ROS, they realized it could revolutionize how they build space robots. Instead of starting from scratch for each mission, they could leverage the work of the entire robotics community. ROS allows us to focus on the unique challenges of space robotics while leveraging proven solutions for the common problems. When we're sending a robot to Mars, we don't want to be debugging basic navigation algorithms. We want to be focused on the science. NASA is now using ROS for several high-profile missions. The Astrobee robots on the International Space Station run on ROS. The upcoming VIPER mission to search for water on the Moon will use ROS-based software. What's really exciting is that we're not just using ROS, we're contributing back to it. The solutions we develop for space robotics often have applications for terrestrial robotics as well. By sharing our work with the community, we're helping to advance robotics for everyone. The story of ROS shows how open source software can transform an entire field. By creating a common platform that everyone could build on, Willow Garage didn't just advance their own research—they accelerated the development of robotics worldwide. ROS managed to build enough critical mass and enough positive energy and a good culture that actually engaged enough people to play together and to do it in a more open way. I think that is what can create a revolution. It's a revolution where anybody can leap into action because nobody has to start from scratch. What ROS has enabled people to do is, you can form a company that's focused on an actual application with actual users who have a problem to solve, and you can leverage the entire world of robotics community just by using ROS. Suddenly, the robots we thought we were building, the robots that would clean our houses and fetch our drinks, could stand on the shoulders of robot giants. They could reach higher every generation and today they're even reaching for the stars. Okay. This is Terry Fong. I am the chief roboticist at the NASA Ames Research Center in California. With open source robotics, the sky is not the limit. Even NASA benefits. [inaudible]. At The International Space Station, for example, our dreams of domestic robots will soon level up with flying robots. This project called Astrobee handles housekeeping duties in zero gravity—autonomous free-flying robotic crewmates for the astronauts. These new helpers are being developed thanks to open source software. One of the decisions we made early on, when we were developing Astrobee, was to make it easily accessible so that the guest scientists could work with it. We chose to release all of the software for controlling the robots, all the software tools. For example, we have a pretty high fidelity, 3D, dynamic simulator so people can simulate, as they're doing development, how Astrobee will fly around. All of that we released as open source and we put the source code out on GitHub. It doesn't stop there either. In 2023, the VIPER robot will be searching the south pole of the moon for frozen water. Thanks to open source innovations, computers on Earth will be able to send VIPER instructions. Part of the software we're using to do that on the ground is ROS. Fong feels that larger open source communities have actually increased the reliability of the robotics he works with. If it's something that's adopted by a community, it becomes sustainable. It's something that people continually work to improve, which means the quality goes up. But more than that, it's something that you can count on being there because people care about it. Together we've come a long way from Rosie the maid. Our goals for robots have changed wildly in the past few decades. What we imagined as an end point, those domestic robot servants, turned out to be just the starting point for a robot revolution that's rocketing to other worlds. And for Fong that means we can form a new kind of relationship with our robots. We don't have to limit our imaginations, trying to make them into elaborate house cleaners and personal assistants. As we send humans deeper into space, what's needed to make them successful? Going to Mars will be unlike anything we've ever done before. How do you make sure that, say, a habitat where humans are going to live is operational and it's safe and it's secure so that when they get there they don't have to scramble to stay alive. Those of us who are working space robotics fully believe that robots should be part of the solution. A fantasy that, half a century ago, involved Rosie the robot cleaning up your living room has evolved into something so much more. But all that work on basic chores for robots did push the industry forward. Soon, Rosie the robot will be fixing up our home on Mars. You know, even today there's still no robot butler, no robot maid—but maybe that's okay. Maybe you can get your own beer. The robot dream of the 1980s lives on anyway, as a kind of museum collection over at Rick Rowland's place. I've collected anything made by Androbot, the Topo series, the BOB series, the Fred series. There was Andy, have the Hero line- When you look at all those dusty old models, you realize it was the dream that mattered. By dreaming up robots that would help us with ordinary tasks, we've pushed ourselves, inspired ourselves to create robots that would one day help with the extraordinary. ROS and the emergence of open source robotics created the potential for a robot revolution far beyond the gussied up Roombas of the '80s. It's a software revolution, and that's what we're exploring next time in episode two. I'm Saron Yitbarek and this is Command Line Heroes, an original podcast from Red Hat. Follow or subscribe wherever you get your podcasts. Until next time, keep on coding. Hero line, the Hero Jr. We have the RB5X robot and the RB5X robot with arms. A robot that came along a little later, it was called the Newton's SynPet. We have all the Omnibot series from Tomy. We have Robie Juniors, Robie Seniors. We have [Inaudible 00:24:54]-

More about 技術の歴史

未来を想像するには、過去を振り返って、感動と刺激を与えるイノベーションの事例を発見することが重要です。

ブログ投稿

Ping コマンドの基本：テストとトラブルシューティング

2021年 4月 26日  |  Damon Garn

...

Linux

ブログ投稿

Linux システムを Active Directory ドメインに参加させる方法

2020年 10月 13日  |   Edem Afenyo

...

Linux

About the show

Command Line Heroes

During its run from 2018 to 2022, Command Line Heroes shared the epic true stories of developers, programmers, hackers, geeks, and open source rebels, and how they revolutionized the technology landscape. Relive our journey through tech history, and use #CommandLinePod to share your favorite episodes.