A folding machine can repeat a known motion once a towel reaches it in the right position. Getting a loose towel into that position is a different job. The robot has to find a useful place to grasp fabric whose shape changes whenever it is lifted, then spread and present it to the machine. That handoff is the gap sewts says its VELUM system addresses in industrial laundries: feeding towels and other terrycloth into existing folders.
The folder needs a prepared towel
Picture two towels arriving at the same station. One lies almost flat, with an edge easy to reach. The other is bunched, with part of its edge hidden under a fold. The folder’s job is still to fold a towel, but its inlet cannot make these two starting shapes identical. Someone or something upstream must turn each towel into an input the folder can use.
That is why a laundry line may wash, dry, and fold textiles with machines yet still need a person stationed at the folding machine. sewts describes the manual step as feeding the laundry and spreading it out. VELUM targets this preparation and feeding step. Calling it a “robot towel folder” would obscure the boundary: the folder does the folding after the towel is presented.
This distinction also limits the claim. The product page specifies towels and other terrycloth. It does not establish that the same setup handles every garment, sheet, or piece of mixed laundry in the same way.
A wrinkle is not a fixed handle
A rigid box keeps its corners where the robot expects them. A towel does not. A visible bump may be an edge, a folded corner, or fabric lying over more fabric. If a gripper lifts the wrong patch, the towel may hang doubled over instead of opening. A weak grip may let it slip, while gripping a folded patch can leave several layers caught together. These are examples of the manipulation problem, not a report of specific VELUM failures or its exact control sequence.
The difficulty continues after the first grasp. As cloth hangs, gravity changes its shape. Spreading it may reveal an edge that was hidden on the conveyor, while another section can sag or fold back. The robot’s task is therefore more than reaching a point in space: it must keep track of what the fabric has become and present a useful shape to the folder. The product description identifies choosing a wrinkle, grasp and pressure as challenges and says sewts analyzes the textile to guide robot commands. It does not publish enough detail there to infer a particular camera, number of grasps, or success rate.
This is one practical example of deformable-object manipulation. The object changes while the robot acts on it, so the next useful move depends on the fabric’s new shape rather than only on a fixed map of the workstation.
What the real installation shows
There is a documented industrial deployment, with an important date attached. In a press release dated January 25, 2023, sewts said the first VELUM delivery had taken place in early November 2022 and that the system was operating at Greif Textile Mietsysteme in Wolfratshausen, Germany. That is evidence of a named early installation as reported by the company. It does not establish how many systems are operating today or how often any particular towel is handled successfully.
For the everyday question, the useful lesson is at the machine boundary. Folding can be a repeatable downstream operation while preparing a crumpled, shape-changing towel for that operation remains a perception and handling problem. VELUM is an example aimed at that feeder gap, within the terrycloth scope its maker describes.