Tag Archives: rich
#435828 Video Friday: Boston Dynamics’ ...
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!):
RoboBusiness 2019 – October 1-3, 2019 – Santa Clara, Calif., USA
ISRR 2019 – October 6-10, 2019 – Hanoi, Vietnam
Ro-Man 2019 – October 14-18, 2019 – New Delhi, India
Humanoids 2019 – October 15-17, 2019 – Toronto, Canada
ARSO 2019 – October 31-1, 2019 – Beijing, China
ROSCon 2019 – October 31-1, 2019 – Macau
IROS 2019 – November 4-8, 2019 – Macau
Let us know if you have suggestions for next week, and enjoy today’s videos.
You’ve almost certainly seen the new Spot and Atlas videos from Boston Dynamics, if for no other reason than we posted about Spot’s commercial availability earlier this week. But what, are we supposed to NOT include them in Video Friday anyway? Psh! Here you go:
[ Boston Dynamics ]
Eight deadly-looking robots. One Giant Nut trophy. Tonight is the BattleBots season finale, airing on Discovery, 8 p.m. ET, or check your local channels.
[ BattleBots ]
Thanks Trey!
Speaking of battling robots… Having giant robots fight each other is one of those things that sounds really great in theory, but doesn’t work out so well in reality. And sadly, MegaBots is having to deal with reality, which means putting their giant fighting robot up on eBay.
As of Friday afternoon, the current bid is just over $100,000 with a week to go.
[ MegaBots ]
Michigan Engineering has figured out the secret formula to getting 150,000 views on YouTube: drone plus nail gun.
[ Michigan Engineering ]
Michael Burke from the University of Edinburgh writes:
We’ve been learning to scoop grapefruit segments using a PR2, by “feeling” the difference between peel and pulp. We use joint torque measurements to predict the probability that the knife is in the peel or pulp, and use this to apply feedback control to a nominal cutting trajectory learned from human demonstration, so that we remain in a position of maximum uncertainty about which medium we’re cutting. This means we slice along the boundary between the two mediums. It works pretty well!
[ Paper ] via [ Robust Autonomy and Decisions Group ]
Thanks Michael!
Hey look, it’s Jan with eight EMYS robot heads. Hi, Jan! Hi, EMYSes!
[ EMYS ]
We’re putting the KRAKEN Arm through its paces, demonstrating that it can unfold from an Express Rack locker on the International Space Station and access neighboring lockers in NASA’s FabLab system to enable transfer of materials and parts between manufacturing, inspection, and storage stations. The KRAKEN arm will be able to change between multiple ’end effector’ tools such as grippers and inspection sensors – those are in development so they’re not shown in this video.
[ Tethers Unlimited ]
UBTECH’s Alpha Mini Robot with Smart Robot’s “Maatje” software is offering healthcare service to children at Praktijk Intraverte Multidisciplinary Institution in Netherlands.
This institution is using Alpha Mini in counseling children’s behavior. Alpha Mini can move and talk to children and offers games and activities to stimulate and interact with them. Alpha Mini talks, helps and motivates children thereby becoming more flexible in society.
[ UBTECH ]
Some impressive work here from Anusha Nagabandi, Kurt Konoglie, Sergey Levine, Vikash Kumar at Google Brain, training a dexterous multi-fingered hand to do that thing with two balls that I’m really bad at.
Dexterous multi-fingered hands can provide robots with the ability to flexibly perform a wide range of manipulation skills. However, many of the more complex behaviors are also notoriously difficult to control: Performing in-hand object manipulation, executing finger gaits to move objects, and exhibiting precise fine motor skills such as writing, all require finely balancing contact forces, breaking and reestablishing contacts repeatedly, and maintaining control of unactuated objects. In this work, we demonstrate that our method of online planning with deep dynamics models (PDDM) addresses both of these limitations; we show that improvements in learned dynamics models, together with improvements in online model-predictive control, can indeed enable efficient and effective learning of flexible contact-rich dexterous manipulation skills — and that too, on a 24-DoF anthropomorphic hand in the real world, using just 2-4 hours of purely real-world data to learn to simultaneously coordinate multiple free-floating objects.
[ PDDM ]
Thanks Vikash!
CMU’s Ballbot has a deceptively light touch that’s ideal for leading people around.
A paper on this has been submitted to IROS 2019.
[ CMU ]
The Autonomous Robots Lab at the University of Nevada is sharing some of the work they’ve done on path planning and exploration for aerial robots during the DARPA SubT Challenge.
[ Autonomous Robots Lab ]
More proof that anything can be a drone if you staple some motors to it. Even 32 feet of styrofoam insulation.
[ YouTube ]
Whatever you think of military drones, we can all agree that they look cool.
[ Boeing ]
I appreciate the fact that iCub has eyelids, I really do, but sometimes, it ends up looking kinda sleepy in research videos.
[ EPFL LASA ]
Video shows autonomous flight of a lightweight aerial vehicle outdoors and indoors on the campus of Carnegie Mellon University. The vehicle is equipped with limited onboard sensing from a front-facing camera and a proximity sensor. The aerial autonomy is enabled by utilizing a 3D prior map built in Step 1.
[ CMU ]
The Stanford Space Robotics Facility allows researchers to test innovative guidance and navigation algorithms on a realistic frictionless, underactuated system.
[ Stanford ASL ]
In this video, Ian and CP discuss Misty’s many capabilities including robust locomotion, obstacle avoidance, 3D mapping/SLAM, face detection and recognition, sound localization, hardware extensibility, photo and video capture, and programmable personality. They also talk about some of the skills he’s built using these capabilities (and others) and how those skills can be expanded upon by you.
[ Misty Robotics ]
This week’s CMU RI Seminar comes from Aaron Parness at Caltech and NASA JPL, on “Robotic Grippers for Planetary Applications.”
The previous generation of NASA missions to the outer solar system discovered salt water oceans on Europa and Enceladus, each with more liquid water than Earth – compelling targets to look for extraterrestrial life. Closer to home, JAXA and NASA have imaged sky-light entrances to lava tube caves on the Moon more than 100 m in diameter and ESA has characterized the incredibly varied and complex terrain of Comet 67P. While JPL has successfully landed and operated four rovers on the surface of Mars using a 6-wheeled rocker-bogie architecture, future missions will require new mobility architectures for these extreme environments. Unfortunately, the highest value science targets often lie in the terrain that is hardest to access. This talk will explore robotic grippers that enable missions to these extreme terrains through their ability to grip a wide variety of surfaces (shapes, sizes, and geotechnical properties). To prepare for use in space where repair or replacement is not possible, we field-test these grippers and robots in analog extreme terrain on Earth. Many of these systems are enabled by advances in autonomy. The talk will present a rapid overview of my work and a detailed case study of an underactuated rock gripper for deflecting asteroids.
[ CMU ]
Rod Brooks gives some of the best robotics talks ever. He gave this one earlier this week at UC Berkeley, on “Steps Toward Super Intelligence and the Search for a New Path.”
[ UC Berkeley ] Continue reading
#435707 AI Agents Startle Researchers With ...
After 25 million games, the AI agents playing hide-and-seek with each other had mastered four basic game strategies. The researchers expected that part.
After a total of 380 million games, the AI players developed strategies that the researchers didn’t know were possible in the game environment—which the researchers had themselves created. That was the part that surprised the team at OpenAI, a research company based in San Francisco.
The AI players learned everything via a machine learning technique known as reinforcement learning. In this learning method, AI agents start out by taking random actions. Sometimes those random actions produce desired results, which earn them rewards. Via trial-and-error on a massive scale, they can learn sophisticated strategies.
In the context of games, this process can be abetted by having the AI play against another version of itself, ensuring that the opponents will be evenly matched. It also locks the AI into a process of one-upmanship, where any new strategy that emerges forces the opponent to search for a countermeasure. Over time, this “self-play” amounted to what the researchers call an “auto-curriculum.”
According to OpenAI researcher Igor Mordatch, this experiment shows that self-play “is enough for the agents to learn surprising behaviors on their own—it’s like children playing with each other.”
Reinforcement is a hot field of AI research right now. OpenAI’s researchers used the technique when they trained a team of bots to play the video game Dota 2, which squashed a world-champion human team last April. The Alphabet subsidiary DeepMind has used it to triumph in the ancient board game Go and the video game StarCraft.
Aniruddha Kembhavi, a researcher at the Allen Institute for Artificial Intelligence (AI2) in Seattle, says games such as hide-and-seek offer a good way for AI agents to learn “foundational skills.” He worked on a team that taught their AllenAI to play Pictionary with humans, viewing the gameplay as a way for the AI to work on common sense reasoning and communication. “We are, however, quite far away from being able to translate these preliminary findings in highly simplified environments into the real world,” says Kembhavi.
Illustration: OpenAI
AI agents construct a fort during a hide-and-seek game developed by OpenAI.
In OpenAI’s game of hide-and-seek, both the hiders and the seekers received a reward only if they won the game, leaving the AI players to develop their own strategies. Within a simple 3D environment containing walls, blocks, and ramps, the players first learned to run around and chase each other (strategy 1). The hiders next learned to move the blocks around to build forts (2), and then the seekers learned to move the ramps (3), enabling them to jump inside the forts. Then the hiders learned to move all the ramps into their forts before the seekers could use them (4).
The two strategies that surprised the researchers came next. First the seekers learned that they could jump onto a box and “surf” it over to a fort (5), allowing them to jump in—a maneuver that the researchers hadn’t realized was physically possible in the game environment. So as a final countermeasure, the hiders learned to lock all the boxes into place (6) so they weren’t available for use as surfboards.
Illustration: OpenAI
An AI agent uses a nearby box to surf its way into a competitor’s fort.
In this circumstance, having AI agents behave in an unexpected way wasn’t a problem: They found different paths to their rewards, but didn’t cause any trouble. However, you can imagine situations in which the outcome would be rather serious. Robots acting in the real world could do real damage. And then there’s Nick Bostrom’s famous example of a paper clip factory run by an AI, whose goal is to make as many paper clips as possible. As Bostrom told IEEE Spectrum back in 2014, the AI might realize that “human bodies consist of atoms, and those atoms could be used to make some very nice paper clips.”
Bowen Baker, another member of the OpenAI research team, notes that it’s hard to predict all the ways an AI agent will act inside an environment—even a simple one. “Building these environments is hard,” he says. “The agents will come up with these unexpected behaviors, which will be a safety problem down the road when you put them in more complex environments.”
AI researcher Katja Hofmann at Microsoft Research Cambridge, in England, has seen a lot of gameplay by AI agents: She started a competition that uses Minecraft as the playing field. She says the emergent behavior seen in this game, and in prior experiments by other researchers, shows that games can be a useful for studies of safe and responsible AI.
“I find demonstrations like this, in games and game-like settings, a great way to explore the capabilities and limitations of existing approaches in a safe environment,” says Hofmann. “Results like these will help us develop a better understanding on how to validate and debug reinforcement learning systems–a crucial step on the path towards real-world applications.”
Baker says there’s also a hopeful takeaway from the surprises in the hide-and-seek experiment. “If you put these agents into a rich enough environment they will find strategies that we never knew were possible,” he says. “Maybe they can solve problems that we can’t imagine solutions to.” Continue reading