Tag Archives: looking

#437643 Video Friday: Matternet Launches Urban ...

Video Friday is your weekly selection of awesome robotics videos, collected by your Automaton bloggers. We’ll also be posting a weekly calendar of upcoming robotics events for the next few months; here's what we have so far (send us your events!):

IROS 2020 – October 25-25, 2020 – [Online]
Bay Area Robotics Symposium – November 20, 2020 – [Online]
ACRA 2020 – December 8-10, 2020 – [Online]
Let us know if you have suggestions for next week, and enjoy today's videos.

Sixteen teams chose their roster of virtual robots and sensor payloads, some based on real-life subterranean robots, and submitted autonomy and mapping algorithms that SubT Challenge officials then tested across eight cave courses in the cloud-based SubT Simulator. Their robots traversed the cave environments autonomously, without any input or adjustments from human operators. The Cave Circuit Virtual Competition teams earned points by correctly finding, identifying, and localizing up to 20 artifacts hidden in the cave courses within five-meter accuracy.

[ SubT ]

This year, the KUKA Innovation Award’s international jury of experts received a total of more than 40 ideas. The five finalist teams had time until November to implement their ideas. A KUKA LBR Med lightweight robot – the first robotic component to be certified for integration into a medical device – has been made available to them for this purpose. Beyond this, the teams have received a training for the hardware and coaching from KUKA experts throughout the competition. At virtual.MEDICA from 16-19.11.2020, the finalists presented their concepts to an international audience of experts and to the Innovation Award jury.

The winner of the KUKA Innovation Award 2020, worth 20,000 euros, is Team HIFUSK from the Scuola Superiore Sant'Anna in Italy.

[ KUKA Innovation Award ]

Like everything else the in-person Cybathlon event was cancelled, but the competition itself took place, just a little more distributed than it would have been otherwise.

[ Cybathlon ]

Matternet, developer of the world's leading urban drone logistics platform, today announced the launch of operations at Labor Berlin Charité Vivantes in Germany. The program kicked-off November 17, 2020 with permanent operations expected to take flight next year, creating the first urban BVLOS [Beyond Visual Line of Sight] medical drone delivery network in the European Union. The drone network expects to significantly improve the timeliness and efficiency of Labor Berlin’s diagnostics services by providing an option to avoid roadway delays, which will improve patient experience with potentially life-saving benefits and lower costs.

Routine BVLOS over an urban area? Impressive.

[ Matternet ]

Robots playing diabolo!

Thanks Thilo!

[ OMRON Sinic X]

Anki's tech has been repackaged into this robot that serves butter:

[ Butter Robot ]

Berkshire Grey just announced our Picking With Purpose Program in which we’ve partnered our robotic automation solutions with food rescue organizations City Harvest and The Greater Boston Food Bank to pick, pack, and distribute food to families in need in time for Thanksgiving. Berkshire Grey donated about 40,000 pounds of food, used one of our robotic automation systems to pick and pack that food into meal boxes for families in need, and our team members volunteered to run the system. City Harvest and The Greater Boston Food Bank are distributing the 4,000 meal boxes we produced. This is just the beginning. We are building a sponsorship program to make Picking With Purpose an ongoing initiative.

[ Berkshire Grey ]

Thanks Peter!

We posted a video previously of Cassie learning to skip, but here's a much more detailed look (accompanying an ICRA submission) that includes some very impressive stair descending.

[ DRL ]

From garage inventors to university students and entrepreneurs, NASA is looking for ideas on how to excavate the Moon’s icy regolith, or dirt, and deliver it to a hypothetical processing plant at the lunar South Pole. The NASA Break the Ice Lunar Challenge, a NASA Centennial Challenge, is now open for registration. The competition will take place over two phases and will reward new ideas and approaches for a system architecture capable of excavating and moving icy regolith and water on the lunar surface.

[ NASA ]

Adaptation to various scene configurations and object properties, stability and dexterity in robotic grasping manipulation is far from explored. This work presents an origami-based shape morphing fingertip design to actively tackle the grasping stability and dexterity problems. The proposed fingertip utilizes origami as its skeleton providing degrees of freedom at desired positions and motor-driven four-bar-linkages as its transmission components to achieve a compact size of the fingertip.

[ Paper ]

“If Roboy crashes… you die.”

[ Roboy ]

Traditionally lunar landers, as well as other large space exploration vehicles, are powered by solar arrays or small nuclear reactors. Rovers and small robots, however, are not big enough to carry their own dedicated power supplies and must be tethered to their larger counterparts via electrical cables. Tethering severely restricts mobility, and cables are prone to failure due to lunar dust (regolith) interfering with electrical contact points. Additionally, as robots become smaller and more complex, they are fitted with additional sensors that require more power, further exacerbating the problem. Lastly, solar arrays are not viable for charging during the lunar night. WiBotic is developing rapid charging systems and energy monitoring base stations for lunar robots, including the CubeRover – a shoebox-sized robot designed by Astrobotic – that will operate autonomously and charge wirelessly on the Moon.

[ WiBotic ]

Watching pick and place robots is my therapy.

[ Soft Robotics ]

It's really, really hard to beat liquid fuel for energy storage, as Quaternium demonstrates with their hybrid drone.

[ Quaternium ]

Thanks Gregorio!

State-of-the-art quadrotor simulators have a rigid and highly-specialized structure: either are they really fast, physically accurate, or photo-realistic. In this work, we propose a novel quadrotor simulator: Flightmare.

[ Flightmare ]

Drones that chuck fire-fighting balls into burning buildings, sure!

[ LARICS ]

If you missed ROS World, that's okay, because all of the talks are now online. Here's the opening keynote from Vivian Chu and Diligent robotics, along with a couple fun lightning talks.

[ ROS World 2020 ]

This week's CMU RI Seminar is by Chelsea Finn from Stanford University, on Data Scalability for Robot Learning.

Recent progress in robot learning has demonstrated how robots can acquire complex manipulation skills from perceptual inputs through trial and error, particularly with the use of deep neural networks. Despite these successes, the generalization and versatility of robots across environment conditions, tasks, and objects remains a major challenge. And, unfortunately, our existing algorithms and training set-ups are not prepared to tackle such challenges, which demand large and diverse sets of tasks and experiences. In this talk, I will discuss two central challenges that pertain to data scalability: first, acquiring large datasets of diverse and useful interactions with the world, and second, developing algorithms that can learn from such datasets. Then, I will describe multiple approaches that we might take to rethink our algorithms and data pipelines to serve these goals. This will include algorithms that allow a real robot to explore its environment in a targeted manner with minimal supervision, approaches that can perform robot reinforcement learning with videos of human trial-and-error experience, and visual model-based RL approaches that are not bottlenecked by their capacity to model everything about the world.

[ CMU RI ] Continue reading

Posted in Human Robots

#437635 Toyota Research Demonstrates ...

Over the last several years, Toyota has been putting more muscle into forward-looking robotics research than just about anyone. In addition to the Toyota Research Institute (TRI), there’s that massive 175-acre robot-powered city of the future that Toyota still plans to build next to Mount Fuji. Even Toyota itself acknowledges that it might be crazy, but that’s just how they roll—as TRI CEO Gill Pratt told me a while back, when Toyota decides to do something, they really do go all-in on it.

TRI has been focusing heavily on home robots, which is reflective of the long-term nature of what TRI is trying to do, because home robots are both the place where we’ll need robots the most at the same time as they’re the place where it’s going to be hardest to deploy them. The unpredictable nature of homes, and the fact that homes tend to have squishy fragile people in them, are robot-unfriendly characteristics, but as the population continues to age (an increasingly acute problem in Japan), homes offer an enormous amount of potential for helping us maintain our independence.

Today, Toyota is showing off some of the research that it’s been working on recently, in the form of a virtual reality presentation in lieu of an in-person press event. For journalists, TRI pre-loaded the recording onto a VR headset, which was FedEx’ed to my house. You can watch the entire 40-minute presentation in 360 video on YouTube (or in VR if you have a headset of your own), but if you don’t watch the whole thing, you should at least check out the full-on GLaDOS (with arms) that TRI thinks belongs in your home.

The presentation features an introduction from Gill Pratt, who looks entirely too comfortable embedded inside of one of TRI’s telepresence robots. The event also covers a lot of territory, but the highlight is almost certainly the new hardware that TRI demonstrates.

Soft bubble gripper

Photo: TRI

This is a “soft bubble gripper,” under development at TRI’s Cambridge, Mass., branch. These passively-compliant, air-filled grippers make it easier to grasp many different kinds of objects safely, but the nifty thing is that they’ve got cameras inside of them watching a pattern of dots on the interior of the soft membrane.

When the outside of the bubble makes contact with an object, the bubble deforms, and the deformation of the dot pattern on the inside can be tracked by the camera to determine both directions and magnitudes of forces. This is a concept that we’ve seen elsewhere before, but TRI’s implementation is a clever way of making an inherently safe end effector that can still perform all the sensing you need it to do for relatively complex manipulation tasks.

The bubble gripper was presented at ICRA this year, and you can read the technical paper here.

Ceiling-mounted home robot

Photo: TRI

I don’t know whether robots dangling from the ceiling was somehow sinister pre-Portal, but it sure as heck is for me having played through that game a couple of times, and it’s since been reinforced by AUTO from WALL-E.

The reason that we generally see robots mounted on the floor or on tables or on mobile bases is that we’re bipeds, not bats, and giving a robot access to a human-like workspace is easiest to do if you also give that robot a human-like position and orientation. And if you want to be able to reach stuff high up, you do what TRI did with their previous generation of kitchen manipulator, and just give it the ability to make itself super tall. But TRI is convinced it’s a good place to put our future home robots:

One innovative concept is a “gantry robot” that would descend from an overhead framework to perform tasks such as loading the dishwasher, wiping surfaces, and clearing clutter. By traveling on the ceiling, the robot avoids the problems of navigating household floor clutter and navigating cramped spaces. When not in use, the robot would tuck itself up out of the way. To further investigate this idea, the team has built a laboratory prototype robot that can do all the same tasks as a floor-based mobile robot but with the innovative overhead mobility system.

Another obvious problem with the gantry robot is that you have to install all kinds of stuff in your ceiling for this to work, which makes it very impractical (if not totally impossible) to introduce a system like this into a home that wasn’t built specifically for it. If, however, you do build a home with a robot like this in mind, the animation below from TRI shows how it could be extra useful. Suddenly, stairs are a non-issue. Payload is presumably also a non-issue, since loads can be transferred to the ceiling. Batteries become unnecessary, so the whole robot can be much lighter weight, which in turn makes it safer. Sensors get a fantastic view, and obstacle avoidance becomes trivial.

Robots as “time machines”

Photo: TRI

TRI’s presentation covered more than what we’ve highlighted here—our focus has been on the hardware prototypes, but TRI had more to talk about, including learning through demonstration, scaling learning through simulation, and how TRI has been working with users to figure out what research directions should be explored. It’s all available right now on YouTube, and it’s well worth 40 minutes of your time.

“What we’re really focused on is this principle idea of amplifying, rather than replacing, human beings”
—Gill Pratt, TRI

It’s only been five years since Toyota announced the $1 billion investment that established TRI, and it feels like the progress that’s been made since then has been substantial. It’s not often that vision, resources, and long-term commitment come together like this, and TRI’s emphasis on making life better for people is one of the things that helps to keep us optimistic about the future of robotics.

“What we’re really focused on is this principle idea of amplifying, rather than replacing, human beings,” Gill Pratt told us. “And what it means to amplify a person, particularly as they’re aging—what we’re really trying to do is build a time machine. This may sound fanciful, and of course we can’t build a real time machine, but maybe we can build robotic assistants to make our lives as we age seem as if we are actually using a time machine.” He explains that it doesn’t mean building robots for convenience or to do our jobs for us. “It means building technology that enables us to continue to live and to work and to relate to each other as if we were younger,” he says. “And that’s really what our main goal is.” Continue reading

Posted in Human Robots

#437630 How Toyota Research Envisions the Future ...

Yesterday, the Toyota Research Institute (TRI) showed off some of the projects that it’s been working on recently, including a ceiling-mounted robot that could one day help us with household chores. That system is just one example of how TRI envisions the future of robotics and artificial intelligence. As TRI CEO Gill Pratt told us, the company is focusing on robotics and AI technology for “amplifying, rather than replacing, human beings.” In other words, Toyota wants to develop robots not for convenience or to do our jobs for us, but rather to allow people to continue to live and work independently even as we age.

To better understand Toyota’s vision of robotics 15 to 20 years from now, it’s worth watching the 20-minute video below, which depicts various scenarios “where the application of robotic capabilities is enabling members of an aging society to live full and independent lives in spite of the challenges that getting older brings.” It’s a long video, but it helps explains TRI’s perspective on how robots will collaborate with humans in our daily lives over the next couple of decades.

Those are some interesting conceptual telepresence-controlled bipeds they’ve got running around in that video, right?

For more details, we sent TRI some questions on how it plans to go from concepts like the ones shown in the video to real products that can be deployed in human environments. Below are answers from TRI CEO Gill Pratt, who is also chief scientist for Toyota Motor Corp.; Steffi Paepcke, senior UX designer at TRI; and Max Bajracharya, VP of robotics at TRI.

IEEE Spectrum: TRI seems to have a more explicit focus on eventual commercialization than most of the robotics research that we cover. At what point TRI starts to think about things like reliability and cost?

Photo: TRI

Toyota is exploring robots capable of manipulating dishes in a sink and a dishwasher, performing experiments and simulations to make sure that the robots can handle a wide range of conditions.

Gill Pratt: It’s a really interesting question, because the normal way to think about this would be to say, well, both reliability and cost are product development tasks. But actually, we need to think about it at the earliest possible stage with research as well. The hardware that we use in the laboratory for doing experiments, we don’t worry about cost there, or not nearly as much as you’d worry about for a product. However, in terms of what research we do, we very much have to think about, is it possible (if the research is successful) for it to end up in a product that has a reasonable cost. Because if a customer can’t afford what we come up with, maybe it has some academic value but it’s not actually going to make a difference in their quality of life in the real world. So we think about cost very much from the beginning.

The same is true with reliability. Right now, we’re working very hard to make our control techniques robust to wide variations in the environment. For instance, in work that Russ Tedrake is doing with manipulating dishes in a sink and a dishwasher, both in physical testing and in simulation, we’re doing thousands and now millions of different experiments to make sure that we can handle the edge cases and it works over a very wide range of conditions.

A tremendous amount of work that we do is trying to bring robotics out of the age of doing demonstrations. There’s been a history of robotics where for some time, things have not been reliable, so we’d catch the robot succeeding just once and then show that video to the world, and people would get the mis-impression that it worked all of the time. Some researchers have been very good about showing the blooper reel too, to show that some of the time, robots don’t work.

“A tremendous amount of work that we do is trying to bring robotics out of the age of doing demonstrations. There’s been a history of robotics where for some time, things have not been reliable, so we’d catch the robot succeeding just once and then show that video to the world, and people would get the mis-impression that it worked all of the time.”
—Gill Pratt, TRI

In the spirit of sharing things that didn’t work, can you tell us a bit about some of the robots that TRI has had under development that didn’t make it into the demo yesterday because they were abandoned along the way?

Steffi Paepcke: We’re really looking at how we can connect people; it can be hard to stay in touch and see our loved ones as much as we would like to. There have been a few prototypes that we’ve worked on that had to be put on the shelf, at least for the time being. We were exploring how to use light so that people could be ambiently aware of one another across distances. I was very excited about that—the internal name was “glowing orb.” For a variety of reasons, it didn’t work out, but it was really fascinating to investigate different modalities for keeping in touch.

Another prototype we worked on—we found through our research that grocery shopping is obviously an important part of life, and for a lot of older adults, it’s not necessarily the right answer to always have groceries delivered. Getting up and getting out of the house keeps you physically active, and a lot of people prefer to continue doing it themselves. But it can be challenging, especially if you’re purchasing heavy items that you need to transport. We had a prototype that assisted with grocery shopping, but when we pivoted our focus to Japan, we found that the inside of a Japanese home really needs to stay inside, and the outside needs to stay outside, so a robot that traverses both domains is probably not the right fit for a Japanese audience, and those were some really valuable lessons for us.

Photo: TRI

Toyota recently demonstrated a gantry robot that would hang from the ceiling to perform tasks like wiping surfaces and clearing clutter.

I love that TRI is exploring things like the gantry robot both in terms of near-term research and as part of its long-term vision, but is a robot like this actually worth pursuing? Or more generally, what’s the right way to compromise between making an environment robot friendly, and asking humans to make changes to their homes?

Max Bajracharya: We think a lot about the problems that we’re trying to address in a holistic way. We don’t want to just give people a robot, and assume that they’re not going to change anything about their lifestyle. We have a lot of evidence from people who use automated vacuum cleaners that people will adapt to the tools you give them, and they’ll change their lifestyle. So we want to think about what is that trade between changing the environment, and giving people robotic assistance and tools.

We certainly think that there are ways to make the gantry system plausible. The one you saw today is obviously a prototype and does require significant infrastructure. If we’re going to retrofit a home, that isn’t going to be the way to do it. But we still feel like we’re very much in the prototype phase, where we’re trying to understand whether this is worth it to be able to bypass navigation challenges, and coming up with the pros and cons of the gantry system. We’re evaluating whether we think this is the right approach to solving the problem.

To what extent do you think humans should be either directly or indirectly in the loop with home and service robots?

Bajracharya: Our goal is to amplify people, so achieving this is going to require robots to be in a loop with people in some form. One thing we have learned is that using people in a slow loop with robots, such as teaching them or helping them when they make mistakes, gives a robot an important advantage over one that has to do everything perfectly 100 percent of the time. In unstructured human environments, robots are going to encounter corner cases, and are going to need to learn to adapt. People will likely play an important role in helping the robots learn. Continue reading

Posted in Human Robots

#437624 AI-Powered Drone Learns Extreme ...

Quadrotors are among the most agile and dynamic machines ever created. In the hands of a skilled human pilot, they can do some astonishing series of maneuvers. And while autonomous flying robots have been getting better at flying dynamically in real-world environments, they still haven’t demonstrated the same level of agility of manually piloted ones.

Now researchers from the Robotics and Perception Group at the University of Zurich and ETH Zurich, in collaboration with Intel, have developed a neural network training method that “enables an autonomous quadrotor to fly extreme acrobatic maneuvers with only onboard sensing and computation.” Extreme.

There are two notable things here: First, the quadrotor can do these extreme acrobatics outdoors without any kind of external camera or motion-tracking system to help it out (all sensing and computing is onboard). Second, all of the AI training is done in simulation, without the need for an additional simulation-to-real-world (what researchers call “sim-to-real”) transfer step. Usually, a sim-to-real transfer step means putting your quadrotor into one of those aforementioned external tracking systems, so that it doesn’t completely bork itself while trying to reconcile the differences between the simulated world and the real world, where, as the researchers wrote in a paper describing their system, “even tiny mistakes can result in catastrophic outcomes.”

To enable “zero-shot” sim-to-real transfer, the neural net training in simulation uses an expert controller that knows exactly what’s going on to teach a “student controller” that has much less perfect knowledge. That is, the simulated sensory input that the student ends up using as it learns to follow the expert has been abstracted to present the kind of imperfect, imprecise data it’s going to encounter in the real world. This can involve things like abstracting away the image part of the simulation until you’d have no way of telling the difference between abstracted simulation and abstracted reality, which is what allows the system to make that sim-to-real leap.

The simulation environment that the researchers used was Gazebo, slightly modified to better simulate quadrotor physics. Meanwhile, over in reality, a custom 1.5-kilogram quadrotor with a 4:1 thrust to weight ratio performed the physical experiments, using only a Nvidia Jetson TX2 computing board and an Intel RealSense T265, a dual fisheye camera module optimized for V-SLAM. To challenge the learning system, it was trained to perform three acrobatic maneuvers plus a combo of all of them:

Image: University of Zurich/ETH Zurich/Intel

Reference trajectories for acrobatic maneuvers. Top row, from left: Power Loop, Barrel Roll, and Matty Flip. Bottom row: Combo.

All of these maneuvers require high accelerations of up to 3 g’s and careful control, and the Matty Flip is particularly challenging, at least for humans, because the whole thing is done while the drone is flying backwards. Still, after just a few hours of training in simulation, the drone was totally real-world competent at these tricks, and could even extrapolate a little bit to perform maneuvers that it was not explicitly trained on, like doing multiple loops in a row. Where humans still have the advantage over drones is (as you might expect since we’re talking about robots) is quickly reacting to novel or unexpected situations. And when you’re doing this sort of thing outdoors, novel and unexpected situations are everywhere, from a gust of wind to a jealous bird.

For more details, we spoke with Antonio Loquercio from the University of Zurich’s Robotics and Perception Group.

IEEE Spectrum: Can you explain how the abstraction layer interfaces with the simulated sensors to enable effective sim-to-real transfer?

Antonio Loquercio: The abstraction layer applies a specific function to the raw sensor information. Exactly the same function is applied to the real and simulated sensors. The result of the function, which is “abstracted sensor measurements,” makes simulated and real observation of the same scene similar. For example, suppose we have a sequence of simulated and real images. We can very easily tell apart the real from the simulated ones given the difference in rendering. But if we apply the abstraction function of “feature tracks,” which are point correspondences in time, it becomes very difficult to tell which are the simulated and real feature tracks, since point correspondences are independent of the rendering. This applies for humans as well as for neural networks: Training policies on raw images gives low sim-to-real transfer (since images are too different between domains), while training on the abstracted images has high transfer abilities.

How useful is visual input from a camera like the Intel RealSense T265 for state estimation during such aggressive maneuvers? Would using an event camera substantially improve state estimation?

Our end-to-end controller does not require a state estimation module. It shares however some components with traditional state estimation pipelines, specifically the feature extractor and the inertial measurement unit (IMU) pre-processing and integration function. The input of the neural networks are feature tracks and integrated IMU measurements. When looking at images with low features (for example when the camera points to the sky), the neural net will mainly rely on IMU. When more features are available, the network uses to correct the accumulated drift from IMU. Overall, we noticed that for very short maneuvers IMU measurements were sufficient for the task. However, for longer ones, visual information was necessary to successfully address the IMU drift and complete the maneuver. Indeed, visual information reduces the odds of a crash by up to 30 percent in the longest maneuvers. We definitely think that event camera can improve even more the current approach since they could provide valuable visual information during high speed.

“The Matty Flip is probably one of the maneuvers that our approach can do very well … It is super challenging for humans, since they don’t see where they’re going and have problems in estimating their speed. For our approach the maneuver is no problem at all, since we can estimate forward velocities as well as backward velocities.”
—Antonio Loquercio, University of Zurich

You describe being able to train on “maneuvers that stretch the abilities of even expert human pilots.” What are some examples of acrobatics that your drones might be able to do that most human pilots would not be capable of?

The Matty Flip is probably one of the maneuvers that our approach can do very well, but human pilots find very challenging. It basically entails doing a high speed power loop by always looking backward. It is super challenging for humans, since they don’t see where they’re going and have problems in estimating their speed. For our approach the maneuver is no problem at all, since we can estimate forward velocities as well as backward velocities.

What are the limits to the performance of this system?

At the moment the main limitation is the maneuver duration. We never trained a controller that could perform maneuvers longer than 20 seconds. In the future, we plan to address this limitation and train general controllers which can fly in that agile way for significantly longer with relatively small drift. In this way, we could start being competitive against human pilots in drone racing competitions.

Can you talk about how the techniques developed here could be applied beyond drone acrobatics?

The current approach allows us to do acrobatics and agile flight in free space. We are now working to perform agile flight in cluttered environments, which requires a higher degree of understanding of the surrounding with respect to this project. Drone acrobatics is of course only an example application. We selected it because it makes a stress test of the controller performance. However, several other applications which require fast and agile flight can benefit from our approach. Examples are delivery (we want our Amazon packets always faster, don’t we?), search and rescue, or inspection. Going faster allows us to cover more space in less time, saving battery costs. Indeed, agile flight has very similar battery consumption of slow hovering for an autonomous drone.

“Deep Drone Acrobatics,” by Elia Kaufmann, Antonio Loquercio, René Ranftl, Matthias Müller, Vladlen Koltun, and Davide Scaramuzza from the Robotics and Perception Group at the University of Zurich and ETH Zurich, and Intel’s Intelligent Systems Lab, was presented at RSS 2020. Continue reading

Posted in Human Robots

#437620 The Trillion-Transistor Chip That Just ...

The history of computer chips is a thrilling tale of extreme miniaturization.

The smaller, the better is a trend that’s given birth to the digital world as we know it. So, why on earth would you want to reverse course and make chips a lot bigger? Well, while there’s no particularly good reason to have a chip the size of an iPad in an iPad, such a chip may prove to be genius for more specific uses, like artificial intelligence or simulations of the physical world.

At least, that’s what Cerebras, the maker of the biggest computer chip in the world, is hoping.

The Cerebras Wafer-Scale Engine is massive any way you slice it. The chip is 8.5 inches to a side and houses 1.2 trillion transistors. The next biggest chip, NVIDIA’s A100 GPU, measures an inch to a side and has a mere 54 billion transistors. The former is new, largely untested and, so far, one-of-a-kind. The latter is well-loved, mass-produced, and has taken over the world of AI and supercomputing in the last decade.

So can Goliath flip the script on David? Cerebras is on a mission to find out.

Big Chips Beyond AI
When Cerebras first came out of stealth last year, the company said it could significantly speed up the training of deep learning models.

Since then, the WSE has made its way into a handful of supercomputing labs, where the company’s customers are putting it through its paces. One of those labs, the National Energy Technology Laboratory, is looking to see what it can do beyond AI.

So, in a recent trial, researchers pitted the chip—which is housed in an all-in-one system about the size of a dorm room mini-fridge called the CS-1—against a supercomputer in a fluid dynamics simulation. Simulating the movement of fluids is a common supercomputer application useful for solving complex problems like weather forecasting and airplane wing design.

The trial was described in a preprint paper written by a team led by Cerebras’s Michael James and NETL’s Dirk Van Essendelft and presented at the supercomputing conference SC20 this week. The team said the CS-1 completed a simulation of combustion in a power plant roughly 200 times faster than it took the Joule 2.0 supercomputer to do a similar task.

The CS-1 was actually faster-than-real-time. As Cerebrus wrote in a blog post, “It can tell you what is going to happen in the future faster than the laws of physics produce the same result.”

The researchers said the CS-1’s performance couldn’t be matched by any number of CPUs and GPUs. And CEO and cofounder Andrew Feldman told VentureBeat that would be true “no matter how large the supercomputer is.” At a point, scaling a supercomputer like Joule no longer produces better results in this kind of problem. That’s why Joule’s simulation speed peaked at 16,384 cores, a fraction of its total 86,400 cores.

A comparison of the two machines drives the point home. Joule is the 81st fastest supercomputer in the world, takes up dozens of server racks, consumes up to 450 kilowatts of power, and required tens of millions of dollars to build. The CS-1, by comparison, fits in a third of a server rack, consumes 20 kilowatts of power, and sells for a few million dollars.

While the task is niche (but useful) and the problem well-suited to the CS-1, it’s still a pretty stunning result. So how’d they pull it off? It’s all in the design.

Cut the Commute
Computer chips begin life on a big piece of silicon called a wafer. Multiple chips are etched onto the same wafer and then the wafer is cut into individual chips. While the WSE is also etched onto a silicon wafer, the wafer is left intact as a single, operating unit. This wafer-scale chip contains almost 400,000 processing cores. Each core is connected to its own dedicated memory and its four neighboring cores.

Putting that many cores on a single chip and giving them their own memory is why the WSE is bigger; it’s also why, in this case, it’s better.

Most large-scale computing tasks depend on massively parallel processing. Researchers distribute the task among hundreds or thousands of chips. The chips need to work in concert, so they’re in constant communication, shuttling information back and forth. A similar process takes place within each chip, as information moves between processor cores, which are doing the calculations, and shared memory to store the results.

It’s a little like an old-timey company that does all its business on paper.

The company uses couriers to send and collect documents from other branches and archives across town. The couriers know the best routes through the city, but the trips take some minimum amount of time determined by the distance between the branches and archives, the courier’s top speed, and how many other couriers are on the road. In short, distance and traffic slow things down.

Now, imagine the company builds a brand new gleaming skyscraper. Every branch is moved into the new building and every worker gets a small filing cabinet in their office to store documents. Now any document they need can be stored and retrieved in the time it takes to step across the office or down the hall to their neighbor’s office. The information commute has all but disappeared. Everything’s in the same house.

Cerebras’s megachip is a bit like that skyscraper. The way it shuttles information—aided further by its specially tailored compiling software—is far more efficient compared to a traditional supercomputer that needs to network a ton of traditional chips.

Simulating the World as It Unfolds
It’s worth noting the chip can only handle problems small enough to fit on the wafer. But such problems may have quite practical applications because of the machine’s ability to do high-fidelity simulation in real-time. The authors note, for example, the machine should in theory be able to accurately simulate the air flow around a helicopter trying to land on a flight deck and semi-automate the process—something not possible with traditional chips.

Another opportunity, they note, would be to use a simulation as input to train a neural network also residing on the chip. In an intriguing and related example, a Caltech machine learning technique recently proved to be 1,000 times faster at solving the same kind of partial differential equations at play here to simulate fluid dynamics.

They also note that improvements in the chip (and others like it, should they arrive) will push back the limits of what can be accomplished. Already, Cerebras has teased the release of its next-generation chip, which will have 2.6 trillion transistors, 850,00 cores, and more than double the memory.

Of course, it still remains to be seen whether wafer-scale computing really takes off. The idea has been around for decades, but Cerebras is the first to pursue it seriously. Clearly, they believe they’ve solved the problem in a way that’s useful and economical.

Other new architectures are also being pursued in the lab. Memristor-based neuromorphic chips, for example, mimic the brain by putting processing and memory into individual transistor-like components. And of course, quantum computers are in a separate lane, but tackle similar problems.

It could be that one of these technologies eventually rises to rule them all. Or, and this seems just as likely, computing may splinter into a bizarre quilt of radical chips, all stitched together to make the most of each depending on the situation.

Image credit: Cerebras Continue reading

Posted in Human Robots