After 28 years of research and development, Boston Dynamics entered the commercial robot market last year with the launch of Spot. This is a ~$75,000 robotic dog that can march through facilities for remote inspections and, with an additional arm attachment, can even open doors and perform remote manipulation.
Today, Boston Dynamics’ quest for commercialization continues with the announcement of a second commercial robot, “Stretch,” a box-moving bot designed to meet the demands of warehouses and distribution centers. The robot is designed to “go where the work is” in a warehouse, unload trucks, unload shipments from the pallet and eventually build orders. For now, we’re seeing a prototype, but Boston Dynamics hopes companies will start buying Stretch when it’s commercially deployed in 2022.
As Boston Dynamics VP of Product Engineering Kevin Blankespoor told us shortly before Stretch’s launch, the company is going where the customers are. “When we released our first Atlas “Next-gen” video,” said Blankespoor, “there was a part of that video showing Atlas moving boxes, and we got a big response from people in the warehouse space. They wanted that. Atlas would come to work in their warehouse.” Atlas is the company’s all-in-one humanoid research robot and is probably way too expensive to be a commercial product.
Blankespoor continues: “We thought, ‘Well, Atlas is probably a bit complicated to actually work in a warehouse, but we could design a robot that is much simpler and has the same features.'” The result was not just another new one. robot, but an entire warehouse-oriented division within Boston Dynamics, led by Blankespoor.
Getting a grip on Stretch
With a clear demand for warehouse robots, Boston Dynamics began experimenting, first with its “Handle” robot. Handle started life based on a new mobility platform with “wheel legs” – that’s two legs, but with wheels on the bottom instead of feet, allowing for all kinds of agile movements. The first version of Handle was a humanoid robot that could do all kinds of crazy tricks thanks to its wheel legs, such as turning around corners, rolling down stairs and jumping on a table. The video shows Handle lifting a box with its two humanoid arms, but the robot was not yet intended for warehouse work.
For version 2, Handle was “reimagined” from what appeared to be a stunt robot in a warehouse robot, and instead of using two arms, it used a large vacuum gripper to lift boxes. The handle still used the wheel leg concept, but now with a large counterweight at the rear, and along with the long “neck” supporting the vacuum gripper at the front, it looked more like a bird. Blankespoor says Handle could cover many Atlas-in-a-warehouse use cases, but with a third as many connections.
With Handle, Boston Dynamics has even gone as far as experimenting with customers. The warehouse work shown in the Handle version 2 video showed the robot loading and unloading a pallet in a large, open space, and Handle was good at that. The problem with Handle is that sometimes warehouse work has to be done in a limited space, such as unloading a truck, and that’s where the robot had problems.
“It became clear that for Handle maneuvering in a tight space was difficult,” explains Blankespoor. “We were able to get the job done and unload all the boxes, but it actually took too long. Every time Handle picked up a box, it had to roll back to the center of the truck so it could turn without collision, roll forward, and place the box .” When lifting boxes, Handle kept his balance with a large, swinging counterweight in the back, and it sounds like the robot’s big bird butt was constantly in the way. “We knew there was another option,” said Blankespoor. “We’ve been considering something like Stretch for years.”
Stretch is the first Boston Dynamics robot to be “completely purpose-built” for the warehouse, and you can see that much of the nimble bird design has been discarded in favor of a large, hulking industrial robot. We’ll start with the basics: the robot is now simply mounted on a large box, so it’s stable by default and no longer needs active balancing. The robot now weighs 1,200 kg, so there’s no need for a large swinging counterweight when lifting – it won’t tip over. The arm can rotate on top of the base, allowing it to unload boxes from a truck onto a conveyor belt without moving and bumping into anything. As a result, Stretch can unload a truck about five times faster than Handle. Stretch can move up to 800 boxes per hour.
Most warehouses are designed around the 48×40-inch dimensions of a pallet, so Stretch’s base happens to have a 48×40-inch footprint and will fit anywhere a pallet will fit. Wheels in every corner of the box, all with independent controls, allow Stretch to move in any direction, including side to side or spinning in place. The gigantic base also means there’s plenty of room for the battery, enough to power Stretch through an eight-hour shift, or up to 16 hours with “extended range option”.
To where the work is
There are many stationary robotic arms that can move boxes, but arms with fixed infrastructure require costly installation, and you need one for each location where a box needs to be moved. Being mobile means Stretch can do the work of multiple stationary arms, depending on the needs of the warehouse, without the need to redesign or install anything. Blankespoor imagines a typical day in the warehouse for Stretch: “Stretch could spend the morning on the inbound side of the warehouse unloading boxes from trucks. It could be the afternoon in the warehouse aisles, building pallets – they go to retailers or e-commerce centers and he may spend the evening loading boxes back into trucks.”
Stationary arms can be as firm as they need to be, but being mobile means Stretch needs to watch his weight. The Boston Dynamics custom arm design is one-fourth the weight of an industrial arm while still being able to lift its predecessor, with a maximum payload capacity of 50 pounds (23 kg) versus the capacity of 33 pounds (15 kg) ) of Handle. The arm had to be designed so that it could reach over pallets and boxes to the top of the truck, where there is not much space. The robot actually grabs the top row of boxes from the side, as it doesn’t fit between the box and the roof.
The last important part of Stretch is the observation mast, a large tower that sits on the same rotating base as the arm and houses most of the robot’s sensors so that it never gets in the way of the arm. The mast houses both 2D and depth sensors, giving Stretch a high view of the surroundings. For vision, the robot uses Boston Dynamics’ “Pick” software, a collection of machine learning-driven algorithms for detecting and moving boxes, which came to the company through an acquisition of Kinema Systems.
The base of Stretch actually has a modular interface to which you can attach various accessories. For unloading trucks, you can attach a conveyor belt to Stretch, so that the robot can take the conveyor belt with it as it goes deeper into the truck. This means all he has to do is pick up a box, spin it around and drop it to unload faster. There is also a pallet truck attachment so the robot can drag a pallet around while building orders. Additional sensors can also be attached to the base for situational awareness such as additional cameras or lidar, or a barcode reader for input.
Boston Dynamics hopes to sell Stretch in 2022, but it still has some iteration before then. What we’re seeing now is the Stretch prototype, while the “product” version of Stretch will go on sale sometime next year. As Blankespoor explains, “The Stretch product will look something like this, but it’s really been completely redesigned from the ground up. Every part has been reworked for manufacturability for cost reduction, reliability and higher performance. So the Stretch product, we’ll start the building first units of that this summer, and then it will go on sale next year. We’ll start rolling out applications step by step that the product can do. The first thing we’ll do is unload trucks, and then a little bit later we start building pallets.” Blankespoor says the final product will have a few more sensors, such as a lidar on the robot’s face.
“The other thing with the prototype is that our entire software team gets off to a flying start with operating system development, vision and autonomy, and testing on real hardware,” Blankespoor tells us, “so that when you get the product, the systems much more mature.”