As far as I can go out, the Amazon warehouses are very organized, very organized, very arranged, and borrowing the absolute. Everything in Amazon Warehouse (usually) is the place where exactly is supposed to be, which is usually distorted in the Pseudorandom tissue box the size of the shoe box along with a group of other foolishness of nonsense. In one way or another, it turns out that this is the most efficient way in time and time to do things, because)As we wrote beforeYou have to think about a process storage The elements are far from the warehouse as well as a process to choose For them, this involves some concessions in favor of space and speed.
For humans, this is not a big problem. When someone orders something on Amazon, a person can wander around these boxes, pay some things away, then pull the element they are looking for. This is exactly the thing that tends to be terrible robots, because this process is not only a little different at a time, as it is very difficult to determine how humans do so.
As you may expect, Amazon is very hard to do this selection problem. Today in an event in GermanyThe company announced Volcan, an automatic system that can store and choose elements in human speeds (ISH).
The last time we talked with us Harun BarringDirector of Applied Sciences in Amazon robotsand Our conversation focuses on storage– Take the elements in boxes. As part of today’s announcement, Amazon revealed that its robots are now A little faster storage From the ordinary person. But in the context of Stow, there is a limited amount that the robot must really understand about what is already happening in the box. Basically, the Stowing Robot mission is that SQUOOSH is all that is currently in a box to one side as much as possible to provide enough space to install a new element in it. As long as the robot is somewhat eager for not crushing anything, it is relatively directly important, at least compared to the choice.
The options that are taken when storing an element in a box will affect the difficulty of removing this element from this box later – called “Adab Ben”. Amazon tries to learn basket etiquette with artificial intelligence to make a more efficient choice.Amazon
The specific problem of selection, regarding robots, is sensing and manipulating chaos. “It is a task rich in communication normally, and we must plan to contact and respond to it,” says Barnes. It is not enough to solve these problems slowly and carefully, because Amazon Robots It tries to put robots in production, which means that their systems are compared directly to an army of non -reconciliation people who are doing this same task very efficiently.
“There is a new scientific challenge here, which is the determination of the correct element,” Parins explains. The thing that must be understood about identifying the elements in the Amazon warehouse is that there is a file a lot Of them: Something like 400 million unique elements. One floor of the Amazon warehouse can easily contain 15,000 centuries, which is more than a million boxes, and has several hundreds of warehouses. This is a lot of things.
In theory, Amazon knows exactly the elements in each box. Amazon is also known (again, in theory), the weight and dimensions of each of these elements, and may have some images for each of the previous times that the element has been stored or chosen. This is a great starting point for identifying items, but as Parnes notes, “We have a lot of unbeat elements – imagine all the different things that you may get in a brown cardboard box.”
Chaos and communication
As much as a properly challenge is to determine an element in a container that may be stuffed with tumor outside From Ben. It is unparalleled for the devices and programs that humans have to do this task by any robot, which is always a problem, but the real complex factor is to deal with the elements that are collected together in a small texture box. The selection process itself includes more than just extracting-with this element outside the box, then you must reach the selection step, which means dropping it in another box or placing it on a carrier or something like that.
“When we originally started, we assumed that we had to carry the element at a distance after we took it out of the box,” Parins explains. “So we were thinking that we needed to absorb a pinch.” Understanding a pinch is when you hold something between a finger (or fingers) and your thumb, and at least for humans, it is a multi -use and reliable way to seize a wide range of things. But as Parnes notes, for robots in this context, it is more complicated: “Even the absorption of a pinch is not perfect because if the edge of a book, or the end of a plastic bag with something inside it, you will not have control of the element and may be unexpectedly floundering around it.”
https://www.youtube.com/watch?
At some point, Parnes and his team realized that although the element had to move beyond merely outside the box, it was not already necessary to move by choosing the robot itself. Instead, they reached a lifting carrier that puts itself directly outside the box from which it was chosen, so that all the robot should do the element from the box and the carrier. “This does not seem grave at the present time,” Parnes admits, but it is a smart use of devices to simplify the problem of manipulation, and it has a side benefit of allowing the robot to work more efficiently, as the carrier can transfer the element during the start of the arm in the next work.
https://www.youtube.com/watch?
Amazon robots contain various techniques to extract elements from the boxes, using different sweeping devices depending on what to choose. R.It is a kind of final response that the regime chooses and depends on the constipation approach to what the element is, and the place of its presence in the box, and also what it is. It is a complex planning problem that Amazon addresses with artificial intelligence, explains Parnes. “We started building Basic models Among the elements, including properties such as a sponge, the extent of fragility, and whether they tend to stumble on other elements or not. So we are trying to learn these things, which is an early stage for us, but we believe that thinking about the characteristics of elements will be important to reach this level of reliability that we need. “
The reliability should be supernatural for Amazon (and with many other commercial robotic publishing) just because small errors are multiplied by huge publishing operations that lead to an unacceptable amount of tension. There is very very Tail Among the unusual things that Amazon robots may face when trying to extract an element from a box. Even if there is some strange Bin mode that may only appear once in a million choices, this still happens several times a day on the scale where Amazon works. Fortunately for Amazon, they have humans around it, and part of the reason that this automatic system can be effective in production at all is that if the robot stumbles, or even sees a box that knows that it is likely to cause problems, it can only surrender, in directing this appointed element to a human meeting, and moving to the next forum.
The other new technology implemented by Amazon is a kind of modern approach to “Visual“Where the robot sees himself moving and then adjusts his movement based on what he sees. Parrance explains:“ It is an important ability because it allows us to meet problems before it occurs. I think this may be our biggest innovation, and it not only extends our problem, but the problems through robots. “
https://www.youtube.com/watch?
(More) Automated future
The weakness was very clear that Amazon (for the better or worse) did not think about storing it and Choose robots In terms of replacing humans completely. There is a long tail of the elements that need a human touch, and it is frankly difficult to imagine any automated treatment system able to provide at least an accidental humanitarian assistance in an environment like the Amazon warehouse, which manages in one way or another to glorify the organization and chaos at the same time.
These robots that store and choose a direct test in a warehouse from Amazon in Germany during the past year are subject, as they actually show methods that human workers can directly benefit from their existence. For example, Amazon centuries can reach 2.5 meters, which means that human workers need Stepladder to reach the highest boxes and bending to reach the lowest level. If robots are primarily assigned to interact with these boxes, people will help work faster while putting less pressure on their bodies.
Since the robots that have so far managed to keep up with human workers, Parnes tells us that focusing on moving forward will be primarily on improving unstalling: “I think our speed is really good. The thing we focus on now is to get this last part of the reliability, and that the next year of our work will be.” Although it may seem that Amazon is working to improve the very specific cases of use, Parrance repeats that the largest image here uses each of those recent elements with 400 million elements in boxes as a unique opportunity to conduct basic research on rapid and reliable manipulation in complex environments.
“If you can build a flag to deal with high contact and high chaos, we will use it everywhere,” says Barnes. “It will be useful for everything, from warehouses to your home. What we are doing now is just the first problems that force us to develop these capabilities, but I think it is the future of automated manipulation.”
From your site articles
Related articles about the web
