Eureka Robot Software
Caterina Haggins
The Eureka E10s is fairly affordable in the face of other robot mop-vacs; our recommendation for an affordable option is $800, while the E10s is $600. You might think to yourself, why buy a more expensive model then? Why spend more if I don't have to?
It's not a bad vacuum. If you can find it on sale and mostly want it for carpet cleaning, you'll likely be satisfied. I was plenty happy with how it vacuumed my carpet. But the mopping and built-in smarts left something to be desired.
The difference is in the mopping job. Most robot vacuum-mops today have rotating scrubbers or refillable water tanks or self-cleaning tools. Not the E10s, which has the same system as robot mops of yore where you pour water into a canteen in the vacuum that's above the single mop pad. Then the vacuum drags the lightly damp pad around your house to mop your home.
It claims it can rise up, but it was such a minimal rise that the damp mop pad dragged on my rug. It didn't feel soaking wet, but my husband said he could tell it was a little damp. My linoleum floor didn't feel freshly mopped after the damp mop pad swiped over it, either.
The little robot also struggled with floor transitions. The spot where my kitchen linoleum turns into the dining room carpet is my least favorite feature in my current apartment. It's now full of cat food and dried rice that the vacuum pushed there instead of cleaning up. It also got stuck five times in a row on the edge of my rug and carpet where there was an awkward space before the bookshelf began. I finally blocked it off with my toddler's chair.
But even when I leave the base in the same spot, the vacuum forgets my maps over and over again. The E10s also gets upset if I ever dare to move it, which is silly, since I have to clean hair off the brush, put water in the tank, and sometimes help it when it gets itself stuck or lost in my house.
You can save maps in the app, but I often forget, don't resave, and end up paying the price. Since I can't casually remap, it feels like a waste to reclean just to try to get a new map for the next round.
It got stuck more often than seems normal. Some of it was user error, when I forget to clear up some cords, but other times the E10s would be on a mission to clean a spot and get itself stuck in the process. Usually on a transition of some kind, almost always my rug.
Even when it wasn't stuck, it bumped around my home like a blind pet looking for its food bowl, or a moth trying to reach a lightbulb encased in glass. It was kind of cute to watch it try and get between the spokes of my husband's office chair, but annoying to get a notification during every single cleaning that the robot was stuck again, somewhere in my house.
I was relieved to find the Find My Robot feature in the app, so that the E10s would make a noise and help me find it when I would get a notification that the robot was stuck yet again. I knew its trouble spots, but it was easier to find it with the sound alerts.
I have mostly carpet, which I think is what made me still mostly like this vacuum, since it could quietly vacuum my whole home to a pretty satisfying level. I also liked the bagless empty bin, which has a Dyson-like design to it, and the small form factor of the base. But if I bought it because I was expecting to mop, I'd be sorely disappointed. And if I'm not buying it for the mopping, I should have just gotten a cheaper vac-only robot.
The stunning prestidigitation, showcased in the video above, is one of nearly 30 tasks that robots have learned to expertly accomplish thanks to Eureka, which autonomously writes reward algorithms to train bots.
The research paper provides in-depth evaluations of 20 Eureka-trained tasks, based on open-source dexterity benchmarks that require robotic hands to demonstrate a wide range of complex manipulation skills.
With a little luck, a lot of preparation and hard work, the Eureka Robotics Team is headed to Worlds for the second time. They beat 24 other Montana teams in the robot games competition at the State Championships in Bozeman securing their spot in Houston in April.
The small team is successful for many reasons, one of those being the middle school robotics team is a feeder program for the high school. It all comes down to the students who participate in the six-month extracurricular activity.
These people they're all cool because they have the same passions as us and it was really awesome because you don't find a lot of robotic nerds in Eureka," said Team Eureka Robotics programmer Benjamin Cooke.
They're like faces for like sheer terror. Like they were so scared that like our main driver wasn't there. But despite all those challenges, we still performed extremely well, and Jackson stepped in, and he did a phenomenal job," said Lord.
And we told the judges that, they thought wow, that's really interesting because usually, people use pre-made switches. However, we just made these switches out of a screw and a soldered piece of metal. And so they worked really well," said Cooke.
Eureka is a wonderful community. They helped us fundraise our way to worlds and without them, one, we wouldn't have a robotics team and two, we wouldn't have had the opportunity to be there. And so it's just wonderful that we had the support of our community," said Lord.
NVIDIA Research recently unveiled Eureka, a new AI agent that can automatically generate algorithms to train robots to perform complex manual tasks. This breakthrough technology combines large language models like GPT-4 with reinforcement learning to allow robots to rapidly acquire new skills.
As described in NVIDIA's announcement and research paper published this week, Eureka works by leveraging GPT-4's natural language capabilities. The system takes the code of a robotic simulation environment as input context. It then uses this to generate reward functions - the algorithms that enable trial-and-error robot learning through positive reinforcement of desired behaviors.
Crucially, Eureka does this without any human prompting or pre-defined templates. The AI agent is fully autonomous in formulating reward algorithms tailored to the task and robot morphology. Anima Anandkumar, NVIDIA's senior director of AI research and one of the contributors to the Eureka paper, highlighted the prevailing obstacles, stating that reward design, often, is predominantly a "trial-and-error process."
Linxi "Jim" Fan, an integral contributor to the project, described Eureka as an ingenious blend of "large language models and NVIDIA GPU-accelerated simulation technologies." He further projected that Eureka would not only enable intricate robot control but would also revolutionize the creation of physically realistic animations for artists.
In tests across 29 distinct environments with 10 different robot platforms, Eureka consistently produced superior reward programs compared to human experts. On over 80% of tasks, the AI-generated rewards led to over 50% better task performance on average.
NVIDIA provides video demonstrations of Eureka teaching a robotic hand to expertly spin a pen at rapid speeds - on par with human performance. This is a particularly impressive feat requiring complex motor control. The AI has also imparted skills like opening drawers, ball throwing, and scissor handling to various robot designs.
Eureka represents a big step forward for utilizing large language models in robotic control. By closing the gap between high-level planning and low-level motor manipulation, the system expands possibilities for efficiently training robots.
NVIDIA is open-sourcing Eureka's code and benchmarks. This will allow developers to build on the research using tools like the Isaac Gym physics simulator. More intelligent robot learning could soon find its way into industrial and consumer applications.
With inventive applications of AI like Eureka, NVIDIA continues spearheading innovation in robotics and beyond. Eureka provides developers an intriguing new option for imbuing next-gen robots with advanced capabilities.
The six-axis robot arm uses cameras and algorithms to plan its motion and how much force to exert in its grip. The system can mimic the dexterity of human fingers and the visual acuity of human eyes, claimed NTU Singapore.
Pham Quang Cuong, an NTU associate professor who founded Eureka Robotics, said Archimedes is different from other robots currently used in industry that have either high accuracy but low agility (where robots perform the same movements repeatedly), or low accuracy but high agility (such as robots handling packages of different sizes in logistics).
While the robot also takes a few hours to slot delicate optics into a designated tray just like a human operator, the operator can now focus on higher-level tasks after taking three minutes to start the robot on its job, said Pham.
Archimedes uses AI to analyze how many lenses there are and their respective sizes, then an algorithm plans the most efficient way to slot them onto the tray. An alert sounds once the task is complete, so the operator can remove the fully loaded tray, ready for the next manufacturing process.
This robotics research was supported by the Singapore-MIT Alliance for Research and Technology (SMART). Its development from a lab concept into a prototype that is now going out to market is one example of technological translations taking place at the NTU Smart Campus.
The Robot Report is launching the Healthcare Robotics Engineering Forum, which will be on Dec. 9-10 in Santa Clara, Calif. The conference and expo will focus on improving the design, development, and manufacture of next-generation healthcare robots. Learn more about the Healthcare Robotics Engineering Forum, and
The robot can slot lenses and mirrors of different sizes into a custom loading tray, to get them ready for coating. This level of precision could benefit the manufacturers of optical products such as cameras, medical imaging, and eyewear because it eliminates defects in production and improves productivity, said NTU Singapore.
c80f0f1006