Tag Archives: Teaching


Novel system allows robots to learn new skills just by looking at you do it

It may soon be possible to teach a robot any task just by showing it how it’s done — a single time.


Image credits John Greenaway / Flickr.

Researchers at UC Berkeley have developed a way to speed up the education of our silicone-brained friends. In a recently published paper, they report on a new learning algorithm that allows robots to mimic an activity it observed just once on video.

Copy, paste

Training robots today is hard work. Even really simple actions like picking up a cup require paragraphs upon paragraphs of code expressly telling the bot what to do each and every step of the way — a process that is hard, complicated, and probably frustrating for us humans.

There’s work to do even after the code is fully laid out. For example, take assembly line workers. After all the instructions are copy-pasted into their circuits, these bots must undergo a long training process during which they must execute every procedure multiple times. They do so until they can perform the task without making any mistake along the way.

More recently, programmers have created software that allows robots to be programmed just by observing certain tasks. While this is more similar to how we or an animal would learn, it’s still clunky to use — currently, we need to show our robotic friends such training videos thousands of times until they get the hang of it.

The team from UC Berkeley, however, describes a new technique they developed that allows robots to learn a certain action just by observing a human do it a single time.

This technique combines imitation learning with a meta-learning algorithm, the team reports. They christened the resulting system ‘model-agnostic meta-learning’ (MAML). Meta-learning basically means ‘learning to learn’. MAML is a process by which a robot builds on prior experience in order to learn something new. If a robot is shown footage of a human picking up an apple and putting it into a cup, for example, it can gauge what its objective is — putting the apple in the cup. As it learns how to handle these objects, it can expand that knowledge to other similar behaviors. So, for example, if you then go on to show it a video of somebody putting an orange down on a plate, it can recognize the overarching behavior and quickly translates that into the motions it needs to do to carry out the task.

Best of all for all those assembly-line robot trainers out there, the bot doesn’t need to know what an orange or a plate is — it will still perform the required task.

In short, MAML provides a platform that allows a neural network (or a robot) to learn a wide variety of tasks starting with relatively little data. It’s almost the polar opposite of how neural networks work today — which master a single task while drawing on a huge dataset.

The team tested MAML on several robots. After a “single video demonstration from a human”, they note, the robots could successfully perform the shown task. “After meta-learning, the robot can learn to place, push, and pick-and-place new objects using just one video of a human performing the manipulation,” they conclude.

The paper “One-Shot Imitation from Observing Humans via Domain-Adaptive Meta-Learning” has been published in the pre-print journal arXiv.

Robot and human hands.

Robot see, robot do: MIT software allows you to instruct a robot without having to code

Researchers have put together C-LEARN, a system that should allow anyone to teach their robot any task without having to code.

The robot chef from the Easy Living scene in Horizons at EPCOT Center.
Image credits Sam Howzit / Flickr.

Quasi-intelligent robots are already a part of our lives, and someday soon, their full-fledged robotic offspring will be too. But until (or rather, unless) they reach a level of intelligence where we can teach them verbally, as you would a child, instructing a robot will require you to know how to code. Since coding is complicated, more complicated than just doing the dishes yourself, anyway, it’s unlikely that regular people will have much use for robots.

Unless, of course, we could de-code the process of instructing robots. Which is exactly what roboticists at the MIT have done. Called C-LEARN, the system should make the task of instructing your robot as easy as teaching a child. Which is a bit of good-news-bad-news, depending on how you feel about the rise of the machines: good, because we can now have robot friends without learning to code, and bad, because technically the bots can use the system to teach one another.

How to train your bot

So as I’ve said, there’re two ways you can go about it. The first one is to program them, which requires expertise in the field of coding and takes a lot of time. The other is to show the bot what you want it to do by tugging on its limbs or moving digital representations of them around, or just doing the task yourself and having it imitate you. For us muggles the latter is the way to go, but it takes a lot of work to teach a machine even simple movements — and then it can only repeat, not adapt them.

C-LEARN is meant to form a middle road and address the shortcoming of these two methods by arming robots with a knowledge base of simple steps that it can intelligently apply when learning a new task. A human user first helps build up this base by working with the robot. The paper describes how the researchers taught Optimus, a two-armed robot, by using software to simulate the motion of its limbs. Like so:

The researchers described movements such as grasping the top of a cylinder or the side of a block, in different positions, retaking each motion for seven times from each position. The motions varied slightly each time, so the robot can look for underlying patterns in the motions and integrate them into the data bank. If for example the simulated grasper always ended up parallel to the object, the robot would note this position is important in the process and would constrain its future motions to attain this parallelism.

By this point, the robot is very similar to a young child, “that just knows how to reach for something and grasp it,” according to D’Arpino. But starting from this database the robot can learn new, complex tasks following a single demonstration. All you have to do is show it what you want done, then approve or correct its attempt.

Does it work?

Robot and human hands.

To test the system, the researchers taught Optimus four multistep tasks — to pick up a bottle and place it in a bucket, to grab and lift a horizontal tray using both hands, to open a box with one hand and use the other to press a button inside it, and finally to grasp a handled cube with one hand and pull a rod out of it with the other. Optimus was shown how to perform each task once, made 10 attempts at each, and succeeded 37 out of 40 times. Which is pretty good.

The team then went one step further and transferred Optimus’s knowledge base and its understanding of the four tasks to a simulation of Atlas, the bullied bot. It managed to complete all four tasks using the data. When researchers corrupted the data banks by deleting some of the information (such as the constraint to place a grasper parallel to the object), Atlas failed to perform the tasks. Such a system would allow us to confer the models of motion created by one bot with thousands of hours of training and experience to any other robot — anywhere in the world, almost instantly.

D’Arpino is now testing whether having Optimus interact with people for the first time can refine its movement models. Afterward, the team wants to make the robots more flexible in how they apply the rules in their data banks, so that they can adjust their learned behavior to whatever situation they’re faced with.

The goal is to make robots who are able to perform complex, dangerous, or just plain boring tasks with high precision. Applications could include bomb defusal, disaster relief, high-precision manufacturing, and helping sick people with housework.

The findings will be presented later this month at the IEEE International Conference on Robotics and Automation in Singapore.

You can read the full paper “C-LEARN: Learning Geometric Constraints from Demonstrations for Multi-Step Manipulation in Shared Autonomy” here.


Active learning greatly outperforms passive lecturing in classrooms

BORED_classroomMost University professors still rely on passive lectures to get their subject across. A meta-study which analyzed 225 studies found that active teaching – lectures that actively engage students and make the learning experience two-way – improves grades and significantly reduces fail rates. The findings add to an already body of literature that suggests the current dominant teaching model is underperforming and obsolete.

Revising the way education is being transferred

“It’s no longer necessary to prove that active-learning methods are better than traditional lectures,” says Rory Waterman, a chemistry professor at the University of Vermont who is an advocate for active-learning methods and a coorganizer of the Cottrell Scholars Collaborative New Faculty Workshop. “The field can instead focus on which active-learning methods are most effective and how they can be best implemented.”

Scott Freeman, a biology lecturer and education researcher at the University of Washington, Seattle, and colleagues combed through a myriad of studies looking for data that would tell them what kind of impact active learning has. In their paper, the researchers define active learning  as any method that engages students in the process of learning as opposed to passively listening to a lecture. This includes anything from so-called ‘clickers’ – an audience response device which allows lecture attendees to participate in the lecture actively – to the common, yet proven study groups, big or small. The findings suggests that active learning outperforms passive lecturing on all levels – be it chemistry or physics, small or large groups.

On average, score cards improved by one-third of a letter grade. While this might not seem like much, the importance of active learning becomes striking when we look at how it improves student retention rates. Students in traditional lectures are 55% more likely to receive a grade of D or F or to withdraw from a class than are students being taught with active-learning approaches. This tremendous improvement, the researchers write, costs only 10% of the lecture’s time. So just by engaging students for even five minutes during a lecture, a professor can significantly improve his class’ scores and overall learning – statistically speaking, at least.

Susan Singer, director of the Division of Undergraduate Education at the National Science Foundation, believes active learning is most important in science disciplines, where student retention rates are usually lower than other fields.

The study warns, however, that it’s not enough to implement active learning in your class – you have to do it right, too.

“You can goof it up if you don’t do it right,” Freeman explains. He’s witnessed “clicker abuse” in some classes. “There’s a literature on how to use clickers effectively. People have never read any of those papers. They’re just doing it off the cuff. For a scientist or engineer who’s trained to respect evidence and act on it, it’s just horrifying.”

Eventually, Freeman hopes, the study might help educators who still rely on traditional teaching methods to revise their course and migrate to a more engaged method.

“Universities are still over-reliant on lecture-based teaching,” Waterman says, “so helping faculty identify the minimum or first steps they need to take in their classrooms to see these incredible gains in student performance has always seemed to me to be the most practical way to advance student-centered learning.”

Chemical Experiments

Fun and Exciting Chemical Experiments for Teaching and Learning

There’s no better way to foster interest in science and chemistry than seeing it in full, dazzling action. Most of the time, kids and young people wouldn’t really be all that interested in how chemistry works. They wouldn’t be particularly bothered about the different reactions you can get out of two different chemicals interacting with each other. Teaching them the calculations can drive them up the wall with boredom even further. But chemistry, as those of us who know better, can be extremely exciting and well worth getting interested in. And this especially works if you know all the right combinations to make an impact.

So in case your baking-soda-and-vinegar volcano isn’t quite enough to dazzle  them, here are some truly incredible chemistry experiments that’ll get them paying attention at every moment:

Elephant ToothpasteElephant Toothpaste

They can laugh all they like at the funny name, but “Elephant Toothpaste” is a really amazing chemistry experiment that can be done in the lab. You will need:

–        Hydrogen Peroxide Solution (30%, but if you need it higher, you will need to take safety precautions)

–        Saturated Potassium Iodide (KI)

–        Dishwashing detergent

–        Food coloring

This is pretty easy. Pour a small amount (about 50ml) of the H2O2 into a flask. Add a little bit of dishwashing detergent to it, mixing it gently. Add in a few drops of food coloring for effect. Now take a step back, and pour in a little of the KI solution (10ml or so). The result should be an exothermic reaction that causes a huge amount of colored foam to form, which would then spurt up and out of the cylinder.

The Science: The H2O2 decomposes into water, and the oxygen is then catalyzed by the iodide ion, forming the foam because of the dishwashing detergent that catches the oxygen into the bubbles.

Pharoah's SnakePharaoh’s Serpent

After I saw that video of what looked like a monster rising out of a cellphone in a microwave, I called up my old chemistry professor over VoIP (i.e RingCentral) to see if it was really possible. He replied that it was unlikely, but he did know of an experiment that would make something of a similar, and ultimately creepier (and cooler) effect. Warning: If you’re going to attempt this experiment, it has to be in controlled conditions.

Pharaoh’s Serpent needs:

–        Hg(SCN])2 R Mercury(II) thiocyanate

–        Lighter (or a lighting setup that could allow you to light the substance from a distance)

–        Aluminum foil

–        Face masks

Because the substance has mercury and cyanide in the fumes, it’s very important to have this experiment in a controlled area, as well as with everyone taking proper precautions. On top of an aluminum foil, pour a small amount of Hg(SCN)2. Then, using the igniter, simply set the substance aflame and step back so as not to inhale any of the fumes. The result is that after the substance burns into a dark ash, a huge, winding, snake-like solid will start “crawling” right out of it, rising out of the chemical reaction, until all the material is spent.

The Science: The heat source creates a rapid exothermic reaction in the substance, making the coiling solid.

Gummy bear from HellThe Gummy Bear from Hell

What seems like an innocent gummy bear can cause quite a dazzling chemical reaction. For this experiment, you will need:

–        One gummy bear (a red one, if you want effect)

–        A small test tube

–        Molten Potassium Chlorate (KClO3) – make sure the sample is very pure!

With the small amount of molten KClO3 inside the test tube, preferably suspended and the opening pointed away from people, carefully drop one red gummy bear into it. Watch as the gummy bear ignites the molten chlorate, creating what looks like a firework blasting right inside the test tube, spewing white fumes.

The Science: Potassium chlorate is combustible, as you can draw oxygen out of it. This is what is often used in high schools to create oxygen gas for experimentation. It’s important that the sample is pure so there won’t be any explosive accidents. The sugar in the gummy bear reacts with the chlorate, igniting it, and creating the combustible oxygen “firework” effect.

Now these are some experiments worth getting into.