Amazon He has developed a fresh warehouse robot that uses the touch to the back around the shelves to find the right product for shipping to customers.
The robot, called Vulcan, is a significant step towards making robots with smaller fingers compared to people. Further tactile abilities of robots can allow them to be more implementation and production work in the coming years.
Aaron Parness, director of Robotics AI Amazon, who managed the development of Vulcan, explains that touch sensing helps the robot move objects on the shelf and determine what it is. “When you try to hide [or pick] Positions in one of these pods, you cannot really do this task without making contact with other items, “he says.
The Vulcan system consists of a conventional robotic arm with a non -standard appendix reminiscent of a punching spatula to the shelf and a loser to grab items to pull them out.
Vulcan has sensors on several ponds that allow the robot to detect the edge and contours of objects. Parness claims that machine learning is the key to understanding sensor signals, and is also part of the algorithmic loop that controls the way the robot takes action. “The special sauce that we have is the interpretation of the torque software and the way we wrap them in our control loop and to our movement plans,” he says.
Amazon today revealed Vulcan at the implementation center in Hamburg, Germany. The company claims that the robot is already working in this facility, and another at Washington Spokane.
Up-to-date robots will operate on the same line as human collectors and will try to save them from work in the field of grip, grabbing more items from the shelves that are high or low. Objects that the robot decides that he cannot find will be moved to human employees.
“Amazon stores many different products in containers, so burying is necessary to draw a specific object to fill out the order,” says Ken Goldberg, a robotic of the University of California in Berkeley. “Until now it was very difficult, so I’m curious about the new system.”
Goldberg claims that research on robotic touch detection has increased in recent years, with many groups working on joint and surface detection. He added, however, that robots have a path before they could adapt to the tactile abilities of bodily employees and blood. “The human sense of touch is extremely sensitive and complex, with a huge dynamic range,” says Goldberg. “Robots develop quickly, but I would be surprised to see an equivalent man [skin] Sensors over the next five to ten years. “
Robot collaborators
Despite this, Vulcan should lend a hand automate more work performed by people in the enormous Empire of Amazon fulfillment centers. The company has increased automation in recent years thanks to robots with AI capable of grabbing and transporting packages and packed boxes. Bading and recovering items from shelves is one of the more complex jobs for robots and is very dependent on human work.
Parness says that he does not anticipate robots that undertake all work done in the centers of fulfillment of Amazon. “We don’t really believe in 100 % automation or light fulfillment,” he says. “We can achieve 75 percent and have robots cooperating with our employees, and the sum would be greater [than either working alone]. “