One of the things that we learned from the DARPA Robotics Challenge is that it’s useful for robots to have legs to walk, but it’s even more useful for robots to be versatile and adaptable, with multimodal locomotion capabilities that they can deploy depending on the situation. At the DRC, we saw all kinds of different designs, but one of the more unique approaches came from the University of Bonn, in Germany, with their robot Momaro.
Momaro used a “centaur” design, with four legs that had wheels on the bottom (like a wheeled quadruped) coupled to a humanoid upper torso with a head and arms. It was the top-ranked European robot in the DRC, completing an almost perfect run in just 34 minutes. We’ve since been wondering whether the centaur design would inspire other disaster robots, and now we know the answer is yes.
Today, a consortium of European research groups announced a new centaur robot platform called (appropriately enough) Centauro. The robot was built at the Italian Institute of Technology (IIT) by a team led by Nikos Tsagarakis, who heads IIT’s Humanoids and Human Centered Mechatronics Lab, in Genoa. The project is part of the CENTAURO Consortium, funded by the European Commission and coordinated by researchers from the University of Bonn—the same group that developed Momaro.
The goal of CENTAURO is building a “human-robot symbiotic system where a human operator is telepresent with its whole body in a Centaur-like robot, which is capable of robust locomotion and dexterous manipulation in the rough terrain and austere conditions characteristic of disasters.” From the look of things, Centauro shares the same kind of rugged capability we’ve seen in other IIT robots, like WALK-MAN, meaning that this could be a robot that manages to be both a research tool and real-world useful.
At 1.5 meter tall and weighing 93 kilograms, Centauro is both larger and somehow more agile than the video might make it seem. It’s made of lightweight metals like aluminum, magnesium, and titanium, with skins of 3D printed plastic. Inside are a trio of computers to handle perception, control, and motion planning, along with enough batteries to keep Centauro moving around for a solid 2.5 hours.
Half of what makes Centauro a centaur are its legs:
Centauro’s legs incorporate six degrees of freedom, that is, they can realize articulated movements in the environment, by rotating and extending hips, knees and ankles, and control the wheel modules, which are placed at the ankles like rolling "hoofs." The robot can adopt different configurations, such as the typical leg configurations of quadruped robots including both inward and outward knee arrangements, and a spider leg configuration, which can be more beneficial in terms of stability required for manipulating powerful tools. Wheels allow the robot to demonstrate also wheel-based mobility in addition to the articulated locomotion. The wheel is made of an aluminium alloy and its outer layer is over-moulded with an elastomer material, thereby ensuring visco-damped contacts while generating suitable friction when rolling on the ground surfaces.
And the other half is its torso:
The Centauro robot is capable of using human tools to execute manipulation tasks and can demonstrate manipulation strength capacity that is higher than that of the typical human adult. Its lightweight (10.5 kg) arms demonstrate a payload to weight ratio greater than 1:1; thus, the payload capacity of the single arm is approximately 11 kg. Furthermore, its high performance and impact-resilient actuation system permits the robot to perform manipulation tasks that require severe physical interactions without the risk of physical damage to robot components.
The idea is that this combination of quadruped and humanoid optimizes both stability and mobility while allowing a remote human operator to much more intuitively control the robot’s arms and hands, since everything is oriented in a familiar way. As we saw in the DRC, it can take a lot of practice to be a robot driver, and the usefulness and efficiency of robots like these in disaster-type scenarios depends on how intuitive and effective their control interface is.
We’re looking forward to seeing more of how Centauro integrates full autonomy, supervised autonomy, and telepresence control. With hardware this capable, the robot may be limited primarily by what it’s smart enough to do on its own, along with the interface that lets a human control it directly. Hopefully, we’ll start seeing more demos from IIT to get a better sense of all the cool stuff that Centauro will be learning how to do. For now, check out this karate chop:
For more details, we spoke with Nikos Tsagarakis, who led the team at IIT that developed Centauro.
IEEE Spectrum: Can you talk about how Centauro is related to Momaro?
Nikos Tsagarakis: The visual perception system of Centauro follows closely the principles and sensor arrangement that was developed in Momaro. And Centauro’s locomotion has been inspired by Momaro as well. But compared to Momaro, Centauro’s leg kinematics is different, providing more flexibility in adjusting the orientation of the leg end-effector (wheel) with respect to terrain. … Furthermore, the manipulation capabilities and kinematics of the Centauro arms are very different, with improved manipulability and strength. Centauro also has torso mobility that is missing from Momaro.
IIT has developed a lot of expertise working on bipedal robot projects like your WALK-MAN humanoid. Why build a four-legged robot now?
Humanoid robots are certainly the main research line of the group at IIT as we believe in long term this body design will be more effective to act and operate within human infrastructures. The Centauro robot actually makes use of the technologies developed within our humanoid development line, including the actuation and design principles, control systems, etc. It represents an effort to bring these technologies to practical applications faster, as we all know that humanoids will need some extra time to reach the level of performance needed to operate robustly in the real world.
Centauro offers an easier platform to use, providing robust and flexible balancing and mobility skills thanks its its hybrid locomotion principles that combine the wheeled and legged functionality. I should also mention that our humanoid line of development is not abandoned but still continues with the same effort and motivation, and new results will be released in the near future.
What kinds of things will Centauro be able to do that WALK-MAN could not do?
The Centauro platform is equipped with more stable locomotion and balancing skills that provide confidence that the robot will be able to transverse real terrain closer to those in real applications involving disaster response.
The robot has been designed to support higher manipulation loads than WALK-MAN and is more resilient against physical impacts, allowing it to perform tasks that require it to modify the environment, e.g., breaking a blocked door to free the pathway.
The robot dimensions are smaller than WALK-MAN, making it easier to operate within human environment. Despite this reduction in size Centauro is stronger and more resilient than WALK-MAN, e.g., its arms have a payload capacity 60 to 70 percent higher than WALK-MAN.
Our next generation of software architecture and whole-body control tools makes it easier to operate the robot through a high level teleoperation interface. The interface features local autonomous modules with whole-body motion generation and impedance regulation that take care of the robot motions and interaction impedances.
With four legs, and the wheeled feet, will it be able to walk over rubble? How about climbing stairs? And can it do any of the other tasks from the DARPA Robotics Challenge?
Definitely yes. We have already performed these tasks in the lab and these demonstrations will be out later this year with the robot performing these operations outdoors. The combination of wheels and legs makes it possible to implement “terrain following” locomotion, which gives the robot greater ability to cope with uneven terrain.
Is the robot designed to be teleoperated most of the time? What functions will it be able to do autonomously?
High level commands come from the pilot, while whole-body motion generation and impedance regulation are modulated by local shared autonomous controllers.
What was the biggest technical challenge in developing the robot, and how did you solve that problem?
We wanted to design a robot that can demonstrate the physical performance of a typical adult while keeping the robot’s size and weight close to the dimensions and weight of the average adult human. This was not a trivial task and required substantial upgrades on the design principles compared to WALK-MAN. For example, Centauro’s arm weighs half as much as WALK-MAN’s arm while still delivering higher physical strength, power, and robustness.
What do you hope Centauro will accomplish over the next year? What future upgrades are you planning for it?
One goal is to validate the robot’s performance in increasingly more complex locomotion-manipulation tasks that will involve high payloads and harsh physical interactions—in other words, situations resembling those encountered by emergency responders. Another goal is the continuous improvement of the robot task-execution autonomy, as we aim to improve the robot’s overall efficiency and demonstrate faster operation under minimal operator interventions.
[ Project Centauro ]
Evan Ackerman is a senior editor at IEEE Spectrum. Since 2007, he has written over 6,000 articles on robotics and technology. He has a degree in Martian geology and is excellent at playing bagpipes.
Erico Guizzo is the digital product manager at IEEE Spectrum. An IEEE Member, he is an electrical engineer by training and has a master’s degree in science writing from MIT.