By Prof. Wyatt Newman, Ph.D., P.E.
It’s been said that there’s nothing like a little healthy competition to motivate a winner. However, in the world of robotics, recent competitions are making it quite clear that winners are everywhere—in every industry and every part of the globe—thanks to the collaborative power of ROS, otherwise known as the Robot Operating System. In fact, if there’s one thing other than sheer inventive genius that can be credited with spurring the massive robot revolution in process today, it is the power of ROS.
What makes ROS so important? In a word, collaboration. For decades, one promising robotics project after another died in the lab, in large part due to the lack of any sort of standardization of a robotics programming language. Stanford Artificial Intelligence Lab was the first to make headway toward tackling the challenge in 2007, and delivered the fist version of ROS in 2008. When ROS became the standard for the Open Source Robotics Foundation (OSRF) in 2013, programmers finally had the ability to reuse code using robotic “packages” that can be integrated much like LEGO blocks. These packages made it possible for programmers to build on previous advancements rather than starting from scratch. It was a change that dramatically accelerated the pace of innovation in the field and fueled the capabilities of roboticists of every kind—from academic researchers to commercial innovators, and to today’s robotics students, who promise to deliver the next wave of intelligent, agile robots in the decades ahead.
The world of robotics is ripe with competition. Researchers compete to deliver the best new theories and concepts. Programmers compete to create the most powerful applications. Companies compete to be the first to market and to grab the greatest market share. In this spirit of healthy competition, the industry is known for its passionate robotics competitions. In this highly competitive forum, teams battle against one another in high-profile challenges that continuously put new capabilities to the test—and deliver highly encouraging results. Here are just two ROS-focused competitions that are demonstrating the system’s power to continuously elevate the capabilities of robots in the real world:
NIST Agile Robotics Competition
The National Institute of Standards and Technology’s Agile Robotics for Industrial Automation Competition (ARIAC) explores the value of using ROS to enable agile robotics. Each year, competitors from around the world—including independent roboticists, university students, companies, labs, and more—are presented with a new simulation that requires them to utilize the latest advancements in artificial intelligence and robot planning. In 2019, the objective of the simulation was to fulfill kitting orders consisting of a list of specific items, each of which had to be acquired, inspected, accurately placed on a tray, and delivered. Teams were required to build a system using a specific list of components, including 2 robot arms on a common rail, 2 mobile robots, a conveyor belt, 8 storage bins, and various cameras, quality sensors, scanners, and other sensors.
Of course, building a system is one thing; building a system that works in an unpredictable environment is another. In this simulation, orders were received unpredictably as network messages. Once received, the system had to keep track of each order, noting the priority of the order and any updates or changes. Parts that failed inspection or were dropped unpredictably, parts shortages, sudden sensor blackouts, and parts that had to be flipped or acquired from a moving conveyor belt added to the challenge. Additionally, the entire system had to use ROS—for everything from receiving orders to interpreting camera images to controlling robot joints. To qualify, competitors had to prototype their solutions within 6 weeks of the challenge. Both that timeframe and the resulting systems would be unthinkable without the standardization and re-usable software provided by ROS.
Fetch Mobile Manipulation Challenge
Fetch Robotics is a start-up company that makes a mobile manipulator that features a human-like arm and 3-D vision—an ideal combination in scenarios where a robot must navigate between locations to acquire or deliver objects, such as kitting and order fulfillment. Their robots are also viewed as a precursor to domestic robots of the future. In the company’s recent robotics competition, participants had to program one of Fetch’s robots to acquire and deliver five types of parts, one of which required insertion and retrieval from a machining station. There were no fixtures; parts were placed randomly in bins or on table surfaces. The robot had to navigate to different sites, recognize and localize various part types, plan arm and gripper motions to acquire the parts, and deliver completed kits to a depot.
One of the distinguishing features of this competition (pictured above) is that teams are able to test their code remotely on a physical Fetch robot located in San Jose. Remote experimentation time was limited to approximately a dozen hours over 3 weeks, making it essential that code was developed to near completion using pure simulation. Even so, teams were remarkably capable in the live competition at the IEEE International Conference on Robotics and Automation in Montreal last month. The competition did a wonderful job of demonstrating the emerging capabilities of remote collaboration, remote programming, and leveraging realistic simulations for off-line code development. All of these capabilities will help accelerate the growth of robotics.
In a completely different type of contest, DARPA (the Defense Advanced Research Projects Agency) is competing on the global stage with its OFFensive Swarm-Enabled Tactics (OFFSET) program which, according to DARPA, “envisions future small-unit infantry forces using swarms comprising upwards of 250 unmanned aircraft systems (UASs) and/or unmanned ground systems (UGSs) to accomplish diverse missions in complex urban environments.” Here too, ROS plays a key role in enabling and accelerating systems development. With the help of ROS, two defense contractors and over a dozen academic institutions are able to work in concert using standardized interfaces, re-using software, and leveraging tight integration with realistic simulators. It’s just one more example of how ROS enables roboticists to work together toward a common goal.
The role of ROS in accelerating the development of new and innovative robotics applications can’t be overstated. ROS is quickly becoming dominant in industrial robots. It is enabling new robotics start-ups and giving developers the collaborative power they need to deliver highly ambitious projects. According to ROS community metrics, there are now more than 100 distinct types of robots using ROS, and ABI Research projects that “nearly 55% of commercial robots shipped by 2024 will have at least one ROS package onboard.” By making it easier, faster, and cheaper to prototype new robots, and by making them compatible with an increasingly standard infrastructure, ROS is enabling a whole new world of collaborative development. Thanks to a little healthy competition—and the power of ROS—businesses, consumers, and investors alike have a lot to look forward to in the years to come.
Prof. Wyatt Newman, Ph.D. has held visiting appointments at Sandia Labs, NASA, and Princeton, and international appointments at Philips (Eindhoven, The Netherlands), including as a Distinguished Visiting Fellow at U. Edinburgh, and as The Hung Hing Ying Distinguished Visiting Professor at U. of Hong Kong. He has led teams in two DARPA grand challenges—autonomous vehicles and disaster response robotics—and in two recent NIST competitions for agile robotics. His textbook, A Systematic Approach to Learning Robot Programming with ROS, was released by CRC Press in 2017.