Implementation of an Automated Cap Labeling System

---

System Description

The in-house automated system labels up to 450 caps per hour, utilizing machine vision to verify label placement accuracy. The projected annual savings from this implementation range from $40,000 to $50,000.

The system automates the processes of cap orientation, label application, inspection, and sorting, which reduces labor costs and eliminates manual operations. A custom design and the use of open-source hardware kept the cost below $1,000, while providing full process control and operational flexibility. To achieve high label placement precision, the drive mechanism was upgraded with a NEMA 24 geared stepper motor, ensuring stable positioning.

---

Production Need

AlgaeBarn, a company specializing in live aquaculture, manages nine product lines, each requiring a unique cap label. Previously, applying thousands of stickers was done manually, distracting staff from their main duties and causing production delays when pre-labeled caps ran out.

---

Process Automation

The repetitive nature of manual label application made it ideal for automation. The goal was to create an autonomous cell capable of orienting each cap, applying the label, checking the result, and sorting the parts without constant operator involvement.

Commercial labeling machines were available, but their cost and limited process control did not meet production requirements. Many systems assume that after initial setup, label positioning remains accurate, but in practice, after 100–200 cycles, shifts in the roll or backing can affect precision.

To ensure flexibility and future scalability, most of the mechanical system was designed in-house using SOLIDWORKS. Direct hardware costs did not exceed $1,000. Developing the system internally enabled inspection of every cap, rather than relying solely on initial mechanical adjustments.

---

Technological Process

The cycle begins with loading caps into a feed elevator. A sensor monitors the feed track and activates the conveyor as needed. The track provides passive orientation: correctly oriented caps proceed to the labeling station, while inverted ones return to the hopper. This approach avoided the need for complex active orientation mechanisms.

A tray accepts one cap at a time. A KEYENCE laser sensor confirms the presence of a cap and signals the controller, which activates a pneumatic cylinder to move the cap into position. Reed switches confirm the required position before the cycle continues.

The geared stepper motor advances the label roll to the separation edge. As the backing changes direction, the label detaches, and a vacuum applicator picks it up and presses it onto the cap. A reed switch detects the completion of the applicator's movement.

The controller sends an MQTT message to a Raspberry Pi with a camera. An OpenCV program captures an image and evaluates the label's position, measuring the distance between the cap and label edges and checking compliance with the selected tolerance. The inspection result is returned to the controller, after which a robotic arm with a vacuum gripper places the cap either in the container for accepted parts or in the reject bin.

---

Engineering Solutions

The main engineering challenge was precise positioning of the label roll. The initial motor drive responded to stop commands with a delay due to inertia, causing label misalignment. To resolve this, a NEMA 24 geared stepper motor was implemented, allowing the controller to set a repeatable number of steps for each label and stop the motor precisely. This improved process stability and highlighted the importance of millimeter-level accuracy in mass production.

---

Implementation Results

The finished system processes about 450 caps per hour, meeting production needs. The operator can load caps, start the machine, and return to other tasks. If the feed ends or an error occurs, the system stops and sends a notification, preventing the release of defective products.

During the first test, 98 out of 100 caps met the label placement tolerance, with two rejected due to minor deviations. The inspection threshold was set conservatively to ensure high product quality.

The project took about four months, running in parallel with other automation tasks. As a result, regular manual sticker application was eliminated, reducing labor and production downtime. Expected annual savings are $40,000–$50,000.

---

Conclusions

For successful automation, it is recommended to start from actual production needs, assessing time, budget, and expected return in advance, and then designing a solution within those parameters. Small manufacturers do not always need the fastest or most expensive equipment—reliable systems that address specific tasks, provide feedback in case of errors, and ensure consistent product quality are essential.