Amazon has developed a new warehouse robot named Vulcan, designed to locate and retrieve products for customer shipments by utilizing touch technology. This advancement signifies a significant step towards enhancing the tactile capabilities of robots, potentially enabling them to undertake more tasks in fulfillment and manufacturing in the future.
Aaron Parness, Amazon’s Director of Robotics AI who spearheaded Vulcan’s development, explained that touch sensing allows the robot to interact with items on a shelf effectively. He noted, “When performing tasks like stowing or picking, contact with other items is necessary.”
The Vulcan system comprises a conventional robotic arm equipped with a customized spatula-like extension for maneuvering within shelves and a suction device for extracting items. The robot features sensors on multiple joints, allowing it to detect item edges and contours. According to Parness, machine learning plays a crucial role in interpreting sensor signals and is integral to the robot’s action control algorithm.
Amazon unveiled Vulcan at a fulfillment center in Hamburg, Germany, where it is currently operational, alongside another center in Spokane, Washington. These robots work alongside human colleagues, alleviating workers from physically demanding tasks by retrieving items from challenging positions. Any items that Vulcan cannot locate are redirected for human handling.
Ken Goldberg, a roboticist at the University of California, Berkeley, commented on the necessity of rummaging for specific items in Amazon’s diverse inventory. He expressed interest in the system, highlighting recent advancements in robotic touch sensing by various groups. However, he noted that robots still lag behind the complex and highly sensitive human touch capabilities, anticipating it will take five to ten years to match them.
Despite this, Vulcan is expected to further automate tasks traditionally done by humans within Amazon’s expansive fulfillment centers. The company has been augmenting automation through AI-equipped robots that manage various logistical tasks. Parness emphasized that full automation is not anticipated, suggesting a hybrid approach where robots undertake up to 75% of tasks alongside human employees to achieve optimal efficiency.