Tag Archives: Final

#437769 Q&A: Facebook’s CTO Is at War With ...

Posted on November 29, 2020 by Android

Photo: Patricia de Melo Moreira/AFP/Getty Images

Facebook chief technology officer Mike Schroepfer leads the company’s AI and integrity efforts.

Facebook’s challenge is huge. Billions of pieces of content—short and long posts, images, and combinations of the two—are uploaded to the site daily from around the world. And any tiny piece of that—any phrase, image, or video—could contain so-called bad content.

In its early days, Facebook relied on simple computer filters to identify potentially problematic posts by their words, such as those containing profanity. These automatically filtered posts, as well as posts flagged by users as offensive, went to humans for adjudication.

In 2015, Facebook started using artificial intelligence to cull images that contained nudity, illegal goods, and other prohibited content; those images identified as possibly problematic were sent to humans for further review.

By 2016, more offensive photos were reported by Facebook’s AI systems than by Facebook users (and that is still the case).

In 2018, Facebook CEO Mark Zuckerberg made a bold proclamation: He predicted that within five or ten years, Facebook’s AI would not only look for profanity, nudity, and other obvious violations of Facebook’s policies. The tools would also be able to spot bullying, hate speech, and other misuse of the platform, and put an immediate end to them.

Today, automated systems using algorithms developed with AI scan every piece of content between the time when a user completes a post and when it is visible to others on the site—just fractions of a second. In most cases, a violation of Facebook’s standards is clear, and the AI system automatically blocks the post. In other cases, the post goes to human reviewers for a final decision, a workforce that includes 15,000 content reviewers and another 20,000 employees focused on safety and security, operating out of more than 20 facilities around the world.

In the first quarter of this year, Facebook removed or took other action (like appending a warning label) on more than 9.6 million posts involving hate speech, 8.6 million involving child nudity or exploitation, almost 8 million posts involving the sale of drugs, 2.3 million posts involving bullying and harassment, and tens of millions of posts violating other Facebook rules.

Right now, Facebook has more than 1,000 engineers working on further developing and implementing what the company calls “integrity” tools. Using these systems to screen every post that goes up on Facebook, and doing so in milliseconds, is sucking up computing resources. Facebook chief technology officer Mike Schroepfer, who is heading up Facebook’s AI and integrity efforts, spoke with IEEE Spectrum about the team’s progress on building an AI system that detects bad content.

Since that discussion, Facebook’s policies around hate speech have come under increasing scrutiny, with particular attention on divisive posts by political figures. A group of major advertisers in June announced that they would stop advertising on the platform while reviewing the situation, and civil rights groups are putting pressure on others to follow suit until Facebook makes policy changes related to hate speech and groups that promote hate, misinformation, and conspiracies.

Facebook CEO Mark Zuckerberg responded with news that Facebook will widen the category of what it considers hateful content in ads. Now the company prohibits claims that people from a specific race, ethnicity, national origin, religious affiliation, caste, sexual orientation, gender identity, or immigration status are a threat to the physical safety, health, or survival of others. The policy change also aims to better protect immigrants, migrants, refugees, and asylum seekers from ads suggesting these groups are inferior or expressing contempt. Finally, Zuckerberg announced that the company will label some problematic posts by politicians and government officials as content that violates Facebook’s policies.

However, civil rights groups say that’s not enough. And an independent audit released in July also said that Facebook needs to go much further in addressing civil rights concerns and disinformation.

Schroepfer indicated that Facebook’s AI systems are designed to quickly adapt to changes in policy. “I don’t expect considerable technical changes are needed to adjust,” he told Spectrum.

This interview has been edited and condensed for clarity.

IEEE Spectrum: What are the stakes of content moderation? Is this an existential threat to Facebook? And is it critical that you deal well with the issue of election interference this year?

Schroepfer: It’s probably existential; it’s certainly massive. We are devoting a tremendous amount of our attention to it.

The idea that anyone could meddle in an election is deeply disturbing and offensive to all of us here, just as people and citizens of democracies. We don’t want to see that happen anywhere, and certainly not on our watch. So whether it’s important to the company or not, it’s important to us as people. And I feel a similar way on the content-moderation side.

There are not a lot of easy choices here. The only way to prevent people, with certainty, from posting bad things is to not let them post anything. We can take away all voice and just say, “Sorry, the Internet’s too dangerous. No one can use it.” That will certainly get rid of all hate speech online. But I don’t want to end up in that world. And there are variants of that world that various governments are trying to implement, where they get to decide what’s true or not, and you as a person don’t. I don’t want to get there either.

My hope is that we can build a set of tools that make it practical for us to do a good enough job, so that everyone is still excited about the idea that anyone can share what they want, and so that Facebook is a safe and reasonable place for people to operate in.

Spectrum: You joined Facebook in 2008, before AI was part of the company’s toolbox. When did that change? When did you begin to think that AI tools would be useful to Facebook?

Schroepfer: Ten years ago, AI wasn’t commercially practical; the technology just didn’t work very well. In 2012, there was one of those moments that a lot of people point to as the beginning of the current revolution in deep learning and AI. A computer-vision model—a neural network—was trained using what we call supervised training, and it turned out to be better than all the existing models.

Spectrum: How is that training done, and how did computer-vision models come to Facebook?

Image: Facebook

Just Broccoli? Facebook’s image analysis algorithms can tell the difference between marijuana [left] and tempura broccoli [right] better than some humans.

Schroepfer: Say I take a bunch of photos and I have people look at them. If they see a photo of a cat, they put a text label that says cat; if it’s one of a dog, the text label says dog. If you build a big enough data set and feed that to the neural net, it learns how to tell the difference between cats and dogs.

Prior to 2012, it didn’t work very well. And then in 2012, there was this moment where it seemed like, “Oh wow, this technique might work.” And a few years later we were deploying that form of technology to help us detect problematic imagery.

Spectrum: Do your AI systems work equally well on all types of prohibited content?

Schroepfer: Nudity was technically easiest. I don’t need to understand language or culture to understand that this is either a naked human or not. Violence is a much more nuanced problem, so it was harder technically to get it right. And with hate speech, not only do you have to understand the language, it may be very contextual, even tied to recent events. A week before the Christchurch shooting [New Zealand, 2019], saying “I wish you were in the mosque” probably doesn’t mean anything. A week after, that might be a terrible thing to say.

Spectrum: How much progress have you made on hate speech?

Schroepfer: AI, in the first quarter of 2020, proactively detected 88.8 percent of the hate-speech content we removed, up from 80.2 percent in the previous quarter. In the first quarter of 2020, we took action on 9.6 million pieces of content for violating our hate-speech policies.

Image: Facebook

Off Label: Sometimes image analysis isn’t enough to determine whether a picture posted violates the company’s policies. In considering these candy-colored vials of marijuana, for example, the algorithms can look at any accompanying text and, if necessary, comments on the post.

Spectrum: It sounds like you’ve expanded beyond tools that analyze images and are also using AI tools that analyze text.

Schroepfer: AI started off as very siloed. People worked on language, people worked on computer vision, people worked on video. We’ve put these things together—in production, not just as research—into multimodal classifiers.

[Schroepfer shows a photo of a pan of Rice Krispies treats, with text referring to it as a “potent batch”] This is a case in which you have an image, and then you have the text on the post. This looks like Rice Krispies. On its own, this image is fine. You put the text together with it in a bigger model; that can then understand what’s going on. That didn’t work five years ago.

Spectrum: Today, every post that goes up on Facebook is immediately checked by automated systems. Can you explain that process?

Image: Facebook

Bigger Picture: Identifying hate speech is often a matter of context. Either the text or the photo in this post isn’t hateful standing alone, but putting them together tells a different story.

Schroepfer: You upload an image and you write some text underneath it, and the systems look at both the image and the text to try to see which, if any, policies it violates. Those decisions are based on our Community Standards. It will also look at other signals on the posts, like the comments people make.

It happens relatively instantly, though there may be times things happen after the fact. Maybe you uploaded a post that had misinformation in it, and at the time you uploaded it, we didn’t know it was misinformation. The next day we fact-check something and scan again; we may find your post and take it down. As we learn new things, we’re going to go back through and look for violations of what we now know to be a problem. Or, as people comment on your post, we might update our understanding of it. If people are saying, “That’s terrible,” or “That’s mean,” or “That looks fake,” those comments may be an interesting signal.

Spectrum: How is Facebook applying its AI tools to the problem of election interference?

Schroepfer: I would split election interference into two categories. There are times when you’re going after the content, and there are times you’re going after the behavior or the authenticity of the person.

On content, if you’re sharing misinformation, saying, “It’s super Wednesday, not super Tuesday, come vote on Wednesday,” that’s a problem whether you’re an American sitting in California or a foreign actor.

Other times, people create a series of Facebook pages pretending they’re Americans, but they’re really a foreign entity. That is a problem on its own, even if all the content they’re sharing completely meets our Community Standards. The problem there is that you have a foreign government running an information operation.

There, you need different tools. What you’re trying to do is put pieces together, to say, “Wait a second. All of these pages—Martians for Justice, Moonlings for Justice, and Venusians for Justice”—are all run by an administrator with an IP address that’s outside the United States. So they’re all connected, even though they’re pretending to not be connected. That’s a very different problem than me sitting in my office in Menlo Park [Calif.] sharing misinformation.

I’m not going to go into lots of technical detail, because this is an area of adversarial nature. The fundamental problem you’re trying to solve is that there’s one entity coordinating the activity of a bunch of things that look like they’re not all one thing. So this is a series of Instagram accounts, or a series of Facebook pages, or a series of WhatsApp accounts, and they’re pretending to be totally different things. We’re looking for signals that these things are related in some way. And we’re looking through the graph [what Facebook calls its map of relationships between users] to understand the properties of this network.

Spectrum: What cutting-edge AI tools and methods have you been working on lately?

Schroepfer: Supervised learning, with humans setting up the instruction process for the AI systems, is amazingly effective. But it has a very obvious flaw: the speed at which you can develop these things is limited by how fast you can curate the data sets. If you’re dealing in a problem domain where things change rapidly, you have to rebuild a new data set and retrain the whole thing.

Self-supervision is inspired by the way people learn, by the way kids explore the world around them. To get computers to do it themselves, we take a bunch of raw data and build a way for the computer to construct its own tests. For language, you scan a bunch of Web pages, and the computer builds a test where it takes a sentence, eliminates one of the words, and figures out how to predict what word belongs there. And because it created the test, it actually knows the answer. I can use as much raw text as I can find and store because it’s processing everything itself and doesn’t require us to sit down and build the information set. In the last two years there has been a revolution in language understanding as a result of AI self-supervised learning.

Spectrum: What else are you excited about?

Schroepfer: What we’ve been working on over the last few years is multilingual understanding. Usually, when I’m trying to figure out, say, whether something is hate speech or not I have to go through the whole process of training the model in every language. I have to do that one time for every language. When you make a post, the first thing we have to figure out is what language your post is in. “Ah, that’s Spanish. So send it to the Spanish hate-speech model.”

We’ve started to build a multilingual model—one box where you can feed in text in 40 different languages and it determines whether it’s hate speech or not. This is way more effective and easier to deploy.

To geek out for a second, just the idea that you can build a model that understands a concept in multiple languages at once is crazy cool. And it not only works for hate speech, it works for a variety of things.

When we started working on this multilingual model years ago, it performed worse than every single individual model. Now, it not only works as well as the English model, but when you get to the languages where you don’t have enough data, it’s so much better. This rapid progress is very exciting.

Spectrum: How do you move new AI tools from your research labs into operational use?

Schroepfer: Engineers trying to make the next breakthrough will often say, “Cool, I’ve got a new thing and it achieved state-of-the-art results on machine translation.” And we say, “Great. How long does it take to run in production?” They say, “Well, it takes 10 seconds for every sentence to run on a CPU.” And we say, “It’ll eat our whole data center if we deploy that.” So we take that state-of-the-art model and we make it 10 or a hundred or a thousand times more efficient, maybe at the cost of a little bit of accuracy. So it’s not as good as the state-of-the-art version, but it’s something we can actually put into our data centers and run in production.

Spectrum: What’s the role of the humans in the loop? Is it true that Facebook currently employs 35,000 moderators?

Schroepfer: Yes. Right now our goal is not to reduce that. Our goal is to do a better job catching bad content. People often think that the end state will be a fully automated system. I don’t see that world coming anytime soon.

As automated systems get more sophisticated, they take more and more of the grunt work away, freeing up the humans to work on the really gnarly stuff where you have to spend an hour researching.

We also use AI to give our human moderators power tools. Say I spot this new meme that is telling everyone to vote on Wednesday rather than Tuesday. I have a tool in front of me that says, “Find variants of that throughout the system. Find every photo with the same text, find every video that mentions this thing and kill it in one shot.” Rather than, I found this one picture, but then a bunch of other people upload that misinformation in different forms.

Another important aspect of AI is that anything I can do to prevent a person from having to look at terrible things is time well spent. Whether it’s a person employed by us as a moderator or a user of our services, looking at these things is a terrible experience. If I can build systems that take the worst of the worst, the really graphic violence, and deal with that in an automated fashion, that’s worth a lot to me. Continue reading →

Posted in Human Robots

#437763 Peer Review of Scholarly Research Gets ...

Posted on November 28, 2020 by Android

In the world of academics, peer review is considered the only credible validation of scholarly work. Although the process has its detractors, evaluation of academic research by a cohort of contemporaries has endured for over 350 years, with “relatively minor changes.” However, peer review may be set to undergo its biggest revolution ever—the integration of artificial intelligence.

Open-access publisher Frontiers has debuted an AI tool called the Artificial Intelligence Review Assistant (AIRA), which purports to eliminate much of the grunt work associated with peer review. Since the beginning of June 2020, every one of the 11,000-plus submissions Frontiers received has been run through AIRA, which is integrated into its collaborative peer-review platform. This also makes it accessible to external users, accounting for some 100,000 editors, authors, and reviewers. Altogether, this helps “maximize the efficiency of the publishing process and make peer-review more objective,” says Kamila Markram, founder and CEO of Frontiers.

AIRA’s interactive online platform, which is a first of its kind in the industry, has been in development for three years.. It performs three broad functions, explains Daniel Petrariu, director of project management: assessing the quality of the manuscript, assessing quality of peer review, and recommending editors and reviewers. At the initial validation stage, the AI can make up to 20 recommendations and flag potential issues, including language quality, plagiarism, integrity of images, conflicts of interest, and so on. “This happens almost instantly and with [high] accuracy, far beyond the rate at which a human could be expected to complete a similar task,” Markram says.

“We have used a wide variety of machine-learning models for a diverse set of applications, including computer vision, natural language processing, and recommender systems,” says Markram. This includes simple bag-of-words models, as well as more sophisticated deep-learning ones. AIRA also leverages a large knowledge base of publications and authors.

Markram notes that, to address issues of possible AI bias, “We…[build] our own datasets and [design] our own algorithms. We make sure no statistical biases appear in the sampling of training and testing data. For example, when building a model to assess language quality, scientific fields are equally represented so the model isn’t biased toward any specific topic.” Machine- and deep-learning approaches, along with feedback from domain experts, including errors, are captured and used as additional training data. “By regularly re-training, we make sure our models improve in terms of accuracy and stay up-to-date.”

The AI’s job is to flag concerns; humans take the final decisions, says Petrariu. As an example, he cites image manipulation detection—something AI is super-efficient at but is nearly impossible for a human to perform with the same accuracy. “About 10 percent of our flagged images have some sort of problem,” he adds. “[In academic publishing] nobody has done this kind of comprehensive check [using AI] before,” says Petrariu. AIRA, he adds, facilitates Frontiers’ mission to make science open and knowledge accessible to all. Continue reading →

Posted in Human Robots

#437735 Robotic Chameleon Tongue Snatches Nearby ...

Posted on November 27, 2020 by Android

Chameleons may be slow-moving lizards, but their tongues can accelerate at astounding speeds, snatching insects before they have any chance of fleeing. Inspired by this remarkable skill, researchers in South Korea have developed a robotic tongue that springs forth quickly to snatch up nearby items.

They envision the tool, called Snatcher, being used by drones and robots that need to collect items without getting too close to them. “For example, a quadrotor with this manipulator will be able to snatch distant targets, instead of hovering and picking up,” explains Gwang-Pil Jung, a researcher at Seoul National University of Science and Technology (SeoulTech) who co-designed the new device.

There has been other research into robotic chameleon tongues, but what’s unique about Snatcher is that it packs chameleon-tongue fast snatching performance into a form factor that’s portable—the total size is 12 x 8.5 x 8.5 centimeters and it weighs under 120 grams. Still, it’s able to fast snatch up to 30 grams from 80 centimeters away in under 600 milliseconds.

Image: SeoulTech

The fast snatching deployable arm is powered by a wind-up spring attached to a motor (a series elastic actuator) combined with an active clutch. The clutch is what allows the single spring to drive both the shooting and the retracting.

To create Snatcher, Jung and a colleague at SeoulTech, Dong-Jun Lee, set about developing a spring-like device that’s controlled by an active clutch combined with a single series elastic actuator. Powered by a wind-up spring, a steel tapeline—analogous to a chameleon’s tongue—passes through two geared feeders. The clutch is what allows the single spring unwinding in one direction to drive both the shooting and the retracting, by switching a geared wheel between driving the forward feeder or the backward feeder.

The end result is a lightweight snatching device that can retrieve an object 0.8 meters away within 600 milliseconds. Jung notes that some other, existing devices designed for retrieval are capable of accomplishing the task quicker, at about 300 milliseconds, but these designs tend to be bulky. A more detailed description of Snatcher was published July 21 in IEEE Robotics and Automation Letters.

Photo: Dong-Jun Lee and Gwang-Pil Jung/SeoulTech

Snatcher’s relative small size means that it can be installed on a DJI Phantom drone. The researchers want to find out if their system can help make package delivery or retrieval faster and safer.

“Our final goal is to install the Snatcher to a commercial drone and achieve meaningful work, such as grasping packages,” says Jung. One of the challenges they still need to address is how to power the actuation system more efficiently. “To solve this issue, we are finding materials having high energy density.” Another improvement is designing a chameleon tongue-like gripper, replacing the simple hook that’s currently used to pick up objects. “We are planning to make a bi-stable gripper to passively grasp a target object as soon as the gripper contacts the object,” says Jung.

< Back to IEEE Journal Watch Continue reading →

Posted in Human Robots

#437721 Video Friday: Child Robot Learning to ...

Posted on November 26, 2020 by Android

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!):

CLAWAR 2020 – August 24-26, 2020 – [Online Conference]
ICUAS 2020 – September 1-4, 2020 – Athens, Greece
ICRES 2020 – September 28-29, 2020 – Taipei, Taiwan
AUVSI EXPONENTIAL 2020 – October 5-8, 2020 – [Online Conference]
IROS 2020 – October 25-29, 2020 – Las Vegas, Nev., USA
CYBATHLON 2020 – November 13-14, 2020 – [Online Event]
ICSR 2020 – November 14-16, 2020 – Golden, Colo., USA
Let us know if you have suggestions for next week, and enjoy today’s videos.

We first met Ibuki, Hiroshi Ishiguro’s latest humanoid robot, a couple of years ago. A recent video shows how Ishiguro and his team are teaching the robot to express its emotional state through gait and body posture while moving.

This paper presents a subjective evaluation of the emotions of a wheeled mobile humanoid robot expressing emotions during movement by replicating human gait-induced upper body motion. For this purpose, we proposed the robot equipped with a vertical oscillation mechanism that generates such motion by focusing on human center-of-mass trajectory. In the experiment, participants watched videos of the robot’s different emotional gait-induced upper body motions, and assess the type of emotion shown, and their confidence level in their answer.

[ Hiroshi Ishiguro Lab ] via [ RobotStart ]

ICYMI: This is a zinc-air battery made partly of Kevlar that can be used to support weight, not just add to it.

Like biological fat reserves store energy in animals, a new rechargeable zinc battery integrates into the structure of a robot to provide much more energy, a team led by the University of Michigan has shown.

The new battery works by passing hydroxide ions between a zinc electrode and the air side through an electrolyte membrane. That membrane is partly a network of aramid nanofibers—the carbon-based fibers found in Kevlar vests—and a new water-based polymer gel. The gel helps shuttle the hydroxide ions between the electrodes. Made with cheap, abundant and largely nontoxic materials, the battery is more environmentally friendly than those currently in use. The gel and aramid nanofibers will not catch fire if the battery is damaged, unlike the flammable electrolyte in lithium ion batteries. The aramid nanofibers could be upcycled from retired body armor.

[ University of Michigan ]

In what they say is the first large-scale study of the interactions between sound and robotic action, researchers at CMU’s Robotics Institute found that sounds could help a robot differentiate between objects, such as a metal screwdriver and a metal wrench. Hearing also could help robots determine what type of action caused a sound and help them use sounds to predict the physical properties of new objects.

[ CMU ]

Captured on Aug. 11 during the second rehearsal of the OSIRIS-REx mission’s sample collection event, this series of images shows the SamCam imager’s field of view as the NASA spacecraft approaches asteroid Bennu’s surface. The rehearsal brought the spacecraft through the first three maneuvers of the sampling sequence to a point approximately 131 feet (40 meters) above the surface, after which the spacecraft performed a back-away burn.

These images were captured over a 13.5-minute period. The imaging sequence begins at approximately 420 feet (128 meters) above the surface – before the spacecraft executes the “Checkpoint” maneuver – and runs through to the “Matchpoint” maneuver, with the last image taken approximately 144 feet (44 meters) above the surface of Bennu.

[ NASA ]

The DARPA AlphaDogfight Trials Final Event took place yesterday; the livestream is like 5 hours long, but you can skip ahead to 4:39 ish to see the AI winner take on a human F-16 pilot in simulation.

Some things to keep in mind about the result: The AI had perfect situational knowledge while the human pilot had to use eyeballs, and in particular, the AI did very well at lining up its (virtual) gun with the human during fast passing maneuvers, which is the sort of thing that autonomous systems excel at but is not necessarily reflective of better strategy.

[ DARPA ]

Coming soon from Clearpath Robotics!

[ Clearpath ]

This video introduces Preferred Networks’ Hand type A, a tendon-driven robot gripper with passively switchable underactuated surface.

[ Preferred Networks ]

CYBATHLON 2020 will take place on 13 – 14 November 2020 – at the teams’ home bases. They will set up their infrastructure for the competition and film their races. Instead of starting directly next to each other, the pilots will start individually and under the supervision of CYBATHLON officials. From Zurich, the competitions will be broadcast through a new platform in a unique live programme.

[ Cybathlon ]

In this project, we consider the task of autonomous car racing in the top-selling car racing game Gran Turismo Sport. Gran Turismo Sport is known for its detailed physics simulation of various cars and tracks. Our approach makes use of maximum-entropy deep reinforcement learning and a new reward design to train a sensorimotor policy to complete a given race track as fast as possible. We evaluate our approach in three different time trial settings with different cars and tracks. Our results show that the obtained controllers not only beat the built-in non-player character of Gran Turismo Sport, but also outperform the fastest known times in a dataset of personal best lap times of over 50,000 human drivers.

[ UZH ]

With the help of the software pitasc from Fraunhofer IPA, an assembly task is no longer programmed point by point, but workpiece-related. Thus, pitasc adapts the assembly process itself for new product variants with the help of updated parameters.

[ Fraunhofer ]

In this video, a multi-material robot simulator is used to design a shape-changing robot, which is then transferred to physical hardware. The simulated and real robots can use shape change to switch between rolling gaits and inchworm gaits, to locomote in multiple environments.

[ Yale ]

This work presents a novel loco-manipulation control framework for the execution of complex tasks with kinodynamic constraints using mobile manipulators. As a representative example, we consider the handling and re-positioning of pallet jacks in unstructured environments. While these results reveal with a proof-of- concept the effectiveness of the proposed framework, they also demonstrate the high potential of mobile manipulators for relieving human workers from such repetitive and labor intensive tasks. We believe that this extended functionality can contribute to increasing the usability of mobile manipulators in different application scenarios.

[ Paper ] via [ IIT ]

I don’t know why this dinosaur ice cream serving robot needs to blow smoke out of its nose, but I like it.

[ Connected Robotics ] via [ RobotStart ]

Guardian S remote visual inspection and surveillance robots make laying cable runs in confined or hard to reach spaces easy. With advanced maneuverability and the ability to climb vertical, ferrous surfaces, the robot reaches areas that are not always easily accessible.

[ Sarcos ]

Looks like the company that bought Anki is working on an add-on to let cars charge while they drive.

[ Digital Dream Labs ]

Chris Atkeson gives a brief talk for the CMU Robotics Institute orientation.

[ CMU RI ]

A UofT Robotics Seminar, featuring Russ Tedrake from MIT and TRI on “Feedback Control for Manipulation.”

Control theory has an answer for just about everything, but seems to fall short when it comes to closing a feedback loop using a camera, dealing with the dynamics of contact, and reasoning about robustness over the distribution of tasks one might find in the kitchen. Recent examples from RL and imitation learning demonstrate great promise, but don’t leverage the rigorous tools from systems theory. I’d like to discuss why, and describe some recent results of closing feedback loops from pixels for “category-level” robot manipulation.

[ UofT ] Continue reading →

Posted in Human Robots

#437667 17 Teams to Take Part in DARPA’s ...

Posted on November 23, 2020 by Android

Among all of the other in-person events that have been totally wrecked by COVID-19 is the Cave Circuit of the DARPA Subterranean Challenge. DARPA has already hosted the in-person events for the Tunnel and Urban SubT circuits (see our previous coverage here), and the plan had always been for a trio of events representing three uniquely different underground environments in advance of the SubT Finals, which will somehow combine everything into one bonkers course.

While the SubT Urban Circuit event snuck in just under the lockdown wire in late February, DARPA made the difficult (but prudent) decision to cancel the in-person Cave Circuit event. What this means is that there will be no Systems Track Cave competition, which is a serious disappointment—we were very much looking forward to watching teams of robots navigating through an entirely unpredictable natural environment with a lot of verticality. Fortunately, DARPA is still running a Virtual Cave Circuit, and 17 teams will be taking part in this competition featuring a simulated cave environment that’s as dynamic and detailed as DARPA can make it.

From DARPA’s press releases:

DARPA’s Subterranean (SubT) Challenge will host its Cave Circuit Virtual Competition, which focuses on innovative solutions to map, navigate, and search complex, simulated cave environments November 17. Qualified teams have until Oct. 15 to develop and submit software-based solutions for the Cave Circuit via the SubT Virtual Portal, where their technologies will face unknown cave environments in the cloud-based SubT Simulator. Until then, teams can refine their roster of selected virtual robot models, choose sensor payloads, and continue to test autonomy approaches to maximize their score.

The Cave Circuit also introduces new simulation capabilities, including digital twins of Systems Competition robots to choose from, marsupial-style platforms combining air and ground robots, and breadcrumb nodes that can be dropped by robots to serve as communications relays. Each robot configuration has an associated cost, measured in SubT Credits – an in-simulation currency – based on performance characteristics such as speed, mobility, sensing, and battery life.

Each team’s simulated robots must navigate realistic caves, with features including natural terrain and dynamic rock falls, while they search for and locate various artifacts on the course within five meters of accuracy to score points during a 60-minute timed run. A correct report is worth one point. Each course contains 20 artifacts, which means each team has the potential for a maximum score of 20 points. Teams can leverage numerous practice worlds and even build their own worlds using the cave tiles found in the SubT Tech Repo to perfect their approach before they submit one official solution for scoring. The DARPA team will then evaluate the solution on a set of hidden competition scenarios.

Of the 17 qualified teams (you can see all of them here), there are a handful that we’ll quickly point out. Team BARCS, from Michigan Tech, was the winner of the SubT Virtual Urban Circuit, meaning that they may be the team to beat on Cave as well, although the course is likely to be unique enough that things will get interesting. Some Systems Track teams to watch include Coordinated Robotics, CTU-CRAS-NORLAB, MARBLE, NUS SEDS, and Robotika, and there are also a handful of brand new teams as well.

Now, just because there’s no dedicated Cave Circuit for the Systems Track teams, it doesn’t mean that there won’t be a Cave component (perhaps even a significant one) in the final event, which as far as we know is still scheduled to happen in fall of next year. We’ve heard that many of the Systems Track teams have been testing out their robots in caves anyway, and as the virtual event gets closer, we’ll be doing a sort of Virtual Systems Track series that highlights how different teams are doing mock Cave Circuits in caves they’ve found for themselves.

For more, we checked in with DARPA SubT program manager Dr. Timothy H. Chung.

IEEE Spectrum: Was it a difficult decision to cancel the Systems Track for Cave?

Tim Chung: The decision to go virtual only was heart wrenching, because I think DARPA’s role is to offer up opportunities that may be unimaginable for some of our competitors, like opening up a cave-type site for this competition. We crawled and climbed through a number of these sites, and I share the sense of disappointment that both our team and the competitors have that we won’t be able to share all the advances that have been made since the Urban Circuit. But what we’ve been able to do is pour a lot of our energy and the insights that we got from crawling around in those caves into what’s going to be a really great opportunity on the Virtual Competition side. And whether it’s a global pandemic, or just lack of access to physical sites like caves, virtual environments are an opportunity that we want to develop.

“The simulator offers us a chance to look at where things could be … it really allows for us to find where some of those limits are in the technology based only on our imagination.”
—Timothy H. Chung, DARPA

What kind of new features will be included in the Virtual Cave Circuit for this competition?

I’m really excited about these particular features because we’re seeing an opportunity for increased synergy between the physical and the virtual. The first I’d say is that we scanned some of the Systems Track robots using photogrammetry and combined that with some additional models that we got from the systems competitors themselves to turn their systems robots into virtual models. We often talk about the sim to real transfer and how successful we can get a simulation to transfer over to the physical world, but now we’ve taken something from the physical world and made it virtual. We’ve validated the controllers as well as the kinematics of the robots, we’ve iterated with the systems competitors themselves, and now we have these 13 robots (air and ground) in the SubT Tech Repo that now all virtual competitors can take advantage of.

We also have additional robot capability. Those comms bread crumbs are common among many of the competitors, so we’ve adopted that in the virtual world, and now you have comms relay nodes that are baked in to the SubT Simulator—you can have either six or twelve comms nodes that you can drop from a variety of our ground robot platforms. We have the marsupial deployment capability now, so now we have parent ground robots that can be mixed and matched with different child drones to become marsupial pairs.

And this is something I’ve been planning for for a while: we now have the ability to trigger things like rock falls. They still don’t quite look like Indiana Jones with the boulder coming down the corridor, but this comes really close. In addition to it just being an interesting and realistic consideration, we get to really dynamically test and stress the robots’ ability to navigate and recognize that something has changed in the environment and respond to it.

Image: DARPA

DARPA is still running a Virtual Cave Circuit, and 17 teams will be taking part in this competition featuring a simulated cave environment.

No simulation is perfect, so can you talk to us about what kinds of things aren’t being simulated right now? Where does the simulator not match up to reality?

I think that question is foundational to any conversation about simulation. I’ll give you a couple of examples:

We have the ability to represent wholesale damage to a robot, but it’s not at the actuator or component level. So there’s not a reliability model, although I think that would be really interesting to incorporate so that you could do assessments on things like mean time to failure. But if a robot falls off a ledge, it can be disabled by virtue of being too damaged to continue.

With communications, and this is one that’s near and dear not only to my heart but also to all of those that have lived through developing communication systems and robotic systems, we’ve gone through and conducted RF surveys of underground environments to get a better handle on what propagation effects are. There’s a lot of research that has gone into this, and trying to carry through some of that realism, we do have path loss models for RF communications baked into the SubT Simulator. For example, when you drop a bread crumb node, it’s using a path loss model so that it can represent the degradation of signal as you go farther into a cave. Now, we’re not modeling it at the Maxwell equations level, which I think would be awesome, but we’re not quite there yet.

We do have things like battery depletion, sensor degradation to the extent that simulators can degrade sensor inputs, and things like that. It’s just amazing how close we can get in some places, and how far away we still are in others, and I think showing where the limits are of how far you can get simulation is all part and parcel of why SubT Challenge wants to have both System and Virtual tracks. Simulation can be an accelerant, but it’s not going to be the panacea for development and innovation, and I think all the competitors are cognizant those limitations.

One of the most amazing things about the SubT Virtual Track is that all of the robots operate fully autonomously, without the human(s) in the loop that the System Track teams have when they compete. Why make the Virtual Track even more challenging in that way?

I think it’s one of the defining, delineating attributes of the Virtual Track. Our continued vision for the simulation side is that the simulator offers us a chance to look at where things could be, and allows for us to explore things like larger scales, or increased complexity, or types of environments that we can’t physically gain access to—it really allows for us to find where some of those limits are in the technology based only on our imagination, and this is one of the intrinsic values of simulation.

But I think finding a way to incorporate human input, or more generally human factors like teleoperation interfaces and the in-situ stress that you might not be able to recreate in the context of a virtual competition provided a good reason for us to delineate the two competitions, with the Virtual Competition really being about the role of fully autonomous or self-sufficient systems going off and doing their solution without human guidance, while also acknowledging that the real world has conditions that would not necessarily be represented by a fully simulated version. Having said that, I think cognitive engineering still has an incredibly important role to play in human robot interaction.

What do we have to look forward to during the Virtual Competition Showcase?

We have a number of additional features and capabilities that we’ve baked into the simulator that will allow for us to derive some additional insights into our competition runs. Those insights might involve things like the performance of one or more robots in a given scenario, or the impact of the environment on different types of robots, and what I can tease is that this will be an opportunity for us to showcase both the technology and also the excitement of the robots competing in the virtual environment. I’m trying not to give too many spoilers, but we’ll have an opportunity to really get into the details.

Check back as we get closer to the 17 November event for more on the DARPA SubT Challenge. Continue reading →

Posted in Human Robots