Objects
A Battery Plug That Gets a Bath, an Eye Exam and a Retirement Date
SK On proposes a production loop that pulls temporary plugs from battery cells, cleans them, inspects them by camera and decides which ones go back to work.


First, the setting
Rechargeable battery cells are filled with liquid electrolyte through a small port during assembly. Keeping that opening clean between steps matters, so lines use temporary seals and fixtures. Small parts like these usually move on vibratory feeders and conveyors rather than by hand.
Reusing a consumable part on a production line means adding a cleaning step, an inspection step and a rule for when to stop. Machine vision is a common way to check small parts for damage or discoloration. Every such check has to balance throwing away good parts against passing bad ones.
What changes in this design
According to the filing's background, a temporary pin is inserted and later removed during secondary battery manufacturing to keep foreign substances out of the electrolyte injection port. The document says there is a growing need to reuse these pins.
In this proposal, pins move through a loop between two sides of the line: cleaning, camera inspection of shape and color, a reuse decision, and a per-pin cycle count that lets a learning module retire pins before they are inspected again.
If the inspection is reliable, a part that would otherwise be used up can go back to work without carrying dust or contamination into an open battery port. The filing doesn't yet offer measurements showing how reliable the inspection is.
The pin that exists only to be pulled out
This temporary pin has a deliberately short stay in each battery. During secondary battery manufacturing, it plugs the electrolyte injection port to keep foreign substances out, and it comes out once that step is finished. It is a part whose entire career consists of being put in and taken out again.
The applicant is SK On Co., Ltd. of Seoul, and the inventors are Jae Hyeok Yu and Eun Sol Ko. They filed in the US on March 9, 2026, claiming priority to a Korean application filed March 10, 2025. The publication, dated September 10, 2026, notes "an increased need for effective reuse of such a temporary pin" and lays out a loop: remove the pin, clean it, inspect it, put it back.
The loop runs across two sides of the line. A pin used in a first side process, labeled C1, is pulled out, cleaned and inspected, then inserted into a second side process, C2, which runs in the opposite direction. That arrangement is what interests me. The reuse doesn't happen in a separate recycling bay. It sits between two halves of the same production line, so any misjudged pin goes straight back into a battery that's still being built.
Ionizer, mesh plate, then a bath of sound
The cleaner does its work in stages. The pin rests on a mesh plate while an ionizer loosens dust and other particles. Whatever comes off falls through the mesh, passes through a filter unit and is drawn into a collector by an intake fan. The housing around the pin can also act as an ultrasonic cleaner. The filing says the goal is to prevent chemical reactions and to keep foreign substances out of the battery process.
The ionizer suits a small part well. Static is what makes fine dust cling, and once the charge is neutralized, gravity and airflow can carry the dust away. The ultrasonic stage is the heavier tool, and it raises questions the application doesn't answer: what liquid is used, how the pin is dried, and how it avoids carrying a film of residue back to a port that's supposed to stay clean. That is a lot of machinery to keep a small plug from being thrown away, and drying is where I'd want more detail.
Pins travel along an inclined vibratory plate or a conveyor, and a shelter on each side of the cleaner gives them a way in and out. It's ordinary parts handling, and that's a strength. Nothing in the loop needs a person to pick up a pin.
Judging a pin by its shape and its color
After cleaning, the pin sits on a transparent plate while a vision sensor unit examines it from above and below. The camera is looking for two things. A change in shape suggests wear or lost durability. Surface discoloration suggests contact with chemical contaminants. A determination module compares those measurements with reference settings for shape and appearance, then sends the pin either back to the line or out to a discharge unit.
The transparent plate is a neat trick. A camera mounted above can't see the face resting on a table, and on a seal, the hidden face may be the one that matters most. Looking up through the plate shows both sides without flipping the pin.
Using color is the riskier bet. Discoloration is a stand-in for chemical change, and it only works when contamination leaves a visible tint. A pin that has touched electrolyte is not obliged to look guilty. To its credit, the filing says the sensor and measuring device may be redesigned for whatever process uses the pin, which concedes that one camera setup won't suit every job.
A counter that learns when to give up
A counting module tracks how many times each pin has been judged reusable. An artificial-intelligence module stores that count alongside the inspection data and learns an allowable reuse cycle. When a pin reaches its learned final cycle, the module notifies the first supply module. The next time that pin comes off the line, it goes straight to discharge without passing through the cleaner or the camera. According to the filing, the payoff is skipping inspection work that would be wasted.
That leaves an open question: how does the system know which pin is which? Counting cycles for each pin requires recognizing an individual pin every time it returns, and the application doesn't explain how that happens. If pins are tracked as a batch instead, the count becomes an average, and the retirement date belongs to the group rather than to any one pin.
The filing also suggests the accumulated data could help select a shape or material that survives more cycles. In that case, the reuse system would end up shaping the design of the pin itself. It's the most ambitious idea in the document and also the least specified. Retiring pins on a learned schedule also means trusting the model over the camera, a choice that should be backed by data.
What a reused pin still has to prove
The application claims several benefits: a more reliable manufacturing process, less indiscriminate use of resources and lower carbon emissions through reuse. None of them comes with numbers. The filing gives no cycle counts, no inspection error rates and no contamination comparison between new and reused pins. It also doesn't weigh the energy used by the ionizer, the fan and the ultrasonic stage against the pins the system saves. The filing adds that the approach could apply to other disposable components used in a similar way. That's plausible, and it's equally untested.
So the first engineering question I'd ask isn't about the AI. It's about how often the camera wrongly accepts a pin. A pin rejected by mistake costs one pin. A pin accepted by mistake goes back into an open battery port. For me, the case for reuse starts with how rarely that second error happens.
The record, briefly. US20260264120A1, “Temporary Pin Reuse System” A temporary pin reuse system includes modules to remove pins from an electrode assembly, supply them to a cleaning module, determine whether cleaned pins are reusable, and insert reusable pins into a manufacturing process.
More angles on this
- A Stack of Evidence Before the DecisionAnother reading from the same G01N classification family, investigating or analysing materials by determining their chemical or physical properties.
- Bioactive Phytochemicals: a gourd-family extract that must pass a test before it gets a jarAnother reading from the same G01N classification family, investigating or analysing materials by determining their chemical or physical properties.
- CPC B08B: Cleaning in general; prevention of fouling in generalThe official classification family this record was filed into, published in the USPTO Cooperative Patent Classification scheme.
- CPC G01N: Investigating or analysing materials by determining their chemical or physical propertiesThe official classification family this record was filed into, published in the USPTO Cooperative Patent Classification scheme.
- What the Bag's Fill Level Changes UpstreamAnother reading from the neighbouring G01F classification family, measuring volume, volume flow, mass flow or liquid level; metering by volume.
Read around this one:Public patent recordUSPTO published figure