Fetch Robotics introduces Fetch and Freight robots: your warehouse will now be automated.

Just a few months ago, all that we knew about Fetch Robotics was the fact that the main group of Unbounded Robotics, all of whose members had previously worked in Willow Garage, developed not one, but two new robots designed to solve logistic problems. Fetch Robotics now announces Fetch and Freight, a muscular mobile manipulator and a vibrant mobile base designed to automate logistics in places such as warehouses.

A visual video of robots in action and an interview with Fetch Robotics CEO Meloni Wise, where she tells why companies like Amazon and Google are desperate for similar robots:


')

Problem

Fetch and Freight are designed to perform work in warehouse conditions. Companies like Amazon have huge warehouses with shelves full of goods, and when you order something from them, someone has to go and take this product from the shelf (or shelves), put it in the basket, and then send it to packing for transportation. This is just a painful task for the warehouse staff, who ultimately walk from one end of the premises to the other, carrying carts with all the goods you so desperately need and that should be delivered to you in a maximum of two days.

Amazon, of course, has its own fleet of robots that can bring shelves, but using these robots (from Kiva Systems) requires reconfiguration of a warehouse with special racks and the addition of other infrastructure. And even with such robotic warehouses, workers still need to manually pick up goods from the shelves. This is probably why Amazon is trying to figure out how to make robots pick up goods from the shelves . And Amazon is not the only company that needs this technology: Google Express, eBay Now and other e-commerce representatives who depend on warehouses and want their customers to receive their goods as quickly as possible, need robots, even if they are still talking about it. don't know.

Nevertheless, designing robots for work in warehouses is not an easy task. Warehouses are structured environments, which is definitely a plus, but there remains a huge amount of variability in goods that need to be taken, large distances between shelves, compressed lines, people who need to go around, and most importantly, robots should be either more cost-effective than people or increase the amount of work to become more efficient. If your robot does not meet these requirements, it will not take root in the warehouse. Fetch Robotics believes that their solution, consisting of two robots, can solve the problem

Decision

In the process of designing Fetch, we had to decide on the application, and then design a robot that would have the necessary specifications for the application, explains Fetch Robotics CEO Meloni Wise. - Since the presence of multiple passes in warehouses is one of the biggest problems in logistics, we create two robots. The mobile manipulator simply cannot move quickly due to its high center of gravity, so we need a faster and smaller robot to zip through the warehouse and deliver the necessary goods. At some point, robots must move as fast as a human, or even faster.

The system from Fetch Robotics uses a relatively large and functional mobile manipulator (Fetch) to pick up items from the storage shelves, and Freight, the mobile base, acts as a stand-alone basket for goods. Fetch can constantly take goods from the shelves, while Freight can move around the warehouse and collect them. Here is a demo video:



Freight can also be used on its own when it moves behind a person through a warehouse (without using beacons, only vision). In this case, employees do not need to drag carts, and also constantly walk back and forth to deliver the selected product. Instead, they can select products, and several Freight robots will deliver. Fetch Robotics calls this system "Follow Pick":

Honestly, we don’t know why the Follow Pick system is not being implemented, as it can provide material and immediate benefits for employees and companies that hire them. We think the reason is that safe and reliable autonomous navigation has never been combined in a robot that would be cost-effective in this context, but this is one of the key differences, since Fetch and Freight are relatively accessible.

Robots

Let's start with the Fetch mobile pad. The central element is a hand with the possibility of reversal and 7 degrees of mobility, which is capable of lifting 6 kilograms. This is a fairly large amount of payload to handle the vast majority (90-95 percent) of all goods in a typical warehouse. Despite the large payload, Wise says that the robot still has low power for human safety, and its software includes the ability to detect any obstacles and immediately stop the hand.

Located between the arm and the telescopic spine, the Fetch grip range covers the area from the floor to a little less than two meters, which corresponds to the manipulations of a person. The modular capture mechanism is equipped with a standard ISO interface and provides a special Ethernet connection for installing a camera for a close-up view or any other sensor that you find useful.



Although Fetch Robotics do not report how fast Fetch can collect goods at the moment, they claim that their goal (and they are sure that it is attainable) is that Fetch works as fast as a human being.

Fetch is equipped with a PrimeSense 3D sensor in the head, which is a rotating device. There are also many points for mounting additional sensors. Base Fetch includes a charging dock, navigation with a 25-meter range and a laser for avoiding obstacles, as well as a differential drive.

Freight is what you get when you pick up a chainsaw and cut Fetch in half. It has the same differential drive, a 25-meter laser, which is used to detect people and the ability to follow them. As Wise commented earlier, the big advantage Freight has over Fetch is its speed: it now moves easily at a speed of 2 meters per second, although it can reach 3 m / s. This is an approximate speed of jogging.


On top of this, Freight has many attachment points so that you can place shelves or something else on it for better performance of the transport system in the warehouse. Freight comfortably handles goods weighing 70 kg. In fact, it can process more if you are willing to slightly reduce its speed (due to the inertia of the moving body).

Both robots can autonomously charge themselves, and their common dock has a very distinctive shape that helps bots deploy it with lasers. Docks can charge robots in 20-minute series to maximize uptime. During smaller loads, the docks stand on a 3-hour deep charge, which helps maintain battery durability.

From batteries to machinery, Fetch designed both robots for work, not for scientific research. They are designed to be durable. Wise says that the basic minimum lifetime of robots is about 13,000 hours in the worst case of use, when the manipulator will constantly function with a payload of 6 kg. In the case of a more typical use, Fetch is expected to be able to work 16 hours a day, 365 days a year for four years before something breaks in it. A freight that does not have a paddle should last even longer.

Along with the robots themselves, Robotics Fetch will also provide high-level control software that can control many different robots and integrate into existing warehouse systems. Once installed, Fetch Robotics will map, produce semantic labeling, adjust power, and distribute dynamic tasks. All this, of course, based on ROS.

Cost and competition

An important question that arises at the moment concerns the cost of these robots. We would like to give you a final answer. But the problem is that for the implementation of this system in warehouse conditions it is necessary to invest in software and hardware of a system consisting of several robots, therefore, according to Wise, the price for a robot will not be appropriate.


Nevertheless, we were able to accurately determine several facts regarding the cost. Wise said that Fetch will cost "less than $ 100,000," and Freight less than a third of the value of Fetch. Even if we assume that Fetch costs about $ 100,000, compared to other mobile manipulators (like Tiago and RB-1), it is relatively inexpensive, considering its strength, 6 kilograms of payload and a 25-meter laser.

We also heard that the research version of Fetch is in development (hurray!) And more detailed information will be available in the next few weeks.

As for Freight, it is easier to compare it with other mobile bases, primarily with the Adept Lynx platform. We’re not sure how much Lynx costs, but we’re almost certain that it’s more (perhaps a lot more) than $ 33,333.32, and Wise argues that in terms of key capabilities, Freight and Lynx are in essence very, very similar.

Future

Wise is confident that Fetch Robotics will be able to supply robots later this year, as promised in February. We have to point out that this is a crazy idea, considering that the companies are only about eight months old and she finished work on her first robot Fetch in March.

Most likely, we will see Fetch and Freight in pilot programs in real storage conditions over the next few months, but the public will see robots for the first time as research versions on ICRA in Seattle next month.



Fetch Robotics have plans for the commercial production of their two robots. The company currently consists of 13 employees (plus five summer interns) and is trying to grow as fast as possible. Wise commented on their plans for the future:

We assume that robots are able to learn new features, such as bundling, assembling, or other static tasks. We consider robots as a platform that can master new functions.