What is Vulcan Amazon? Types And Vulcan robots price

What is Vulcan Amazon?

Vulcan is the first touch-sensitive robot from Amazon. Vulcan is advancing physical AI, engineering, and robotics to make our workers’ jobs safer and easier while increasing order movement efficiency. The next time you drop a coin on the floor, think about how you will pick it up.

The purpose of Vulcan Amazon robots is to pick and store things from fabric storage pods in Amazon fulfilment centers (FCs). At Amazon’s Delivering the Future summit in Dortmund, Germany, these robots were unveiled. By adding a sensation of touch, they mark a significant advancement in robotics.

Vulcan interacts with fabric pods, as opposed to earlier robotic arms in Amazon FCs that move objects from open-topped bins or conveyor belts. According to one description, a fabric pod is only accessible from the front and resembles a collection of cubbyholes. The contents of these cubbies are stacked, arranged at random, and secured with elastic bands. The Vulcan robots were created to manage the physical interaction with other objects and the pod walls that is frequently required when removing or placing goods in these cubbies.

The Vulcan robots’ primary novelty is their use of touch in addition to vision. Conventional industrial automation, such as other Amazon manipulation efforts, usually avoids physical touch and depends only on shared position information and vision. Due to their inability to sense touch, these “numb and dumb” robots may suddenly stop or smash upon unexpected contact. Vulcan robots, on the other hand, are made especially to come into contact with arbitrary items in their workspace.

Their end-of-arm tools have sensors that monitor torque and force along all six axes. They can use these sensors to measure the force they are applying to objects and take a step back before using too much force. This method is comparable to the way a person locates and retrieves a coin via touch. With the potential to be used for 20 years, this “force-in-the-loop, high-contact style of manipulation” is regarded as a novel paradigm for robotic manipulation.

Vulcan Pick and Vulcan Stow are the two primary robot types included in the Vulcan project.

Vulcan Stow

The purpose of the Vulcan Stow Robot is to store new things in fabric storage pods.

• Item Reception: A computer vision system determines the item’s dimensions to adjust the gripper width after receiving items from a conveyor belt.
• Vision System: Creates an accurate three-dimensional model of the pod and its contents by using three pairs of stereo cameras affixed to a tower.
• Imaging Algorithm: Using simulated bin pictures with elastic bands inserted using a generative-AI model, the algorithm was taught to address the challenge of imaging objects with elastic bands.
• Segmentation: Bin pictures are divided into bins, elastic bands, and objects inside the bands using three deep learning models. A composite 3-D segmentation is produced by projecting these segments onto a 3-D point cloud.
• Space Identification: Determines the vacant space in each bin by computing bounding boxes. The bin is chosen if there is sufficient free space overall. The robot can push objects out of the way if the vacant space is not contiguous.
• Planning Insertion: This method finds insertion points in a two-dimensional image by using convolution with a kernel that represents the necessary space. After projecting this onto the three-dimensional model, a machine learning model creates affordances that show where the object and the extensible blade should be inserted.
• End-Of-Arm Tool (EOAT): A gripper with two parallel plates that slide objects into the bin using conveyor belts. To move objects aside, it also features an extendable aluminium attachment that resembles a kitchen spatula.
• Execution: Based on the affordances, strings collectively control primitives (such as approach, extend blade, sweep, and eject_item). In order to create room, it can rotate things 90 degrees and insert the blade horizontally.
• Justification: In order to calculate the required force, the system takes into account the physics of interacting with various materials (squishy vs. stiff) and forecasts how a pile of things would move when swept.

Vulcan Pick robot

• Operation: Selects certain objects from fabric storage pods.
• Eligibility Check: An item is initially assessed to see if it can be extracted robotically; if not, it is passed to human pickers. FC pictures enhanced with metadata are used in this check.
• Vision System: Measures depth using structured light, or an infrared pattern, from an integrated camera on its EOAT.
• Segmentation and classification: Segmentation is done using a single MaskDINO neural model. Segmented items are categorized into four groups by an additional layer: not an item, item in excellent status, item below others, and item obstructed by others.
  The segmented image is projected onto a point cloud in 3D scene representation. characterizes the three-dimensional scene using a signed distance function, where empty space has a value of positive and occupied space behind a surface has a value of negative.
• Item Identification: Without scanning barcodes, this method instantly compares target photos to sample product photographs to identify the target object. For this, a contrastive learning-trained product-matching model is employed.
• Adhesion Point Selection: Uses the 3-D composite to find comparatively flat regions on the target object that are appropriate for suction. The signed distances of neighbouring regions, which show the probability of collisions, are used to rank these surfaces.
  One type of suction tool is the End-Of-Arm Tool (EOAT).
• Extraction: Applys the suction tool to the surface that has the highest ranking among the candidates. Throughout the extraction process, the suction pressure is tracked. Ten low-resolution pictures are taken by the camera every second to look for variations in bin geometry. It attempts other candidates if the first select point doesn’t work. If there are too many failures, the item is submitted for human removal.

The second arm of the Vulcan Stow and Vulcan Pick robots has a hook EOAT that is used to push up or draw down the elastic bands that cover the bin fronts.

The distribution of products to pods and shelves is purposefully random to increase efficiency and reduce the possibility that many stations would need the same pod at the same time.

These robots’ main goals and effects are as follows: •

• Improving worker safety and ergonomics by managing tasks on the tallest and lowest shelves that would otherwise necessitate the use of stepladders or less ergonomic work positions.
• Allowing human workers to concentrate on center shelves, harder-to-reach goods, or tightly packed containers.
• Increasing the effectiveness of operations.
• Revolutionising operations by making it possible for robots to operate in previously unthinkable situations and objects.

Six Vulcan Stow robots participated in a pilot project in Spokane, Washington, and the robots have successfully finished it. They are now prepared for beta testing, which will take place at the same facility with an additional 30 robots. Vulcan Stow and Vulcan Pick robots will collaborate in a wider deployment at a plant in Germany. They are expected to work at speeds similar to those of front-line staff, handling roughly 75% of the different item kinds kept in FCs.

FAQs

News source