Dual-purpose robotic cell for assembly and ultrasonic deburring.

Problem

Deburring was the actual need. Assembly was there so the robot had something to do when there was no deburring to run — the customer's engineers needed a utilization case strong enough to get the capital approved by their own management, and one process alone did not produce it.

That turns a flexibility requirement into a commercial one, and it changes what "good" means. Idle time is the thing being engineered away, so the changeover between the two modes has to be short enough and reliable enough to actually happen during a production week — not just short enough to demonstrate once at FAT.

Architecture

• Automatic robot tool change enabled the cell to switch between assembly and deburring tasks.
• Servo and pneumatic grippers handled different part and process requirements.
• A four-camera vision system supported part location, validation, and process decisions.
• Modbus-controlled feeders supplied parts into the automated workflow.
• A regrip station allowed the robot to change part orientation for downstream operations.
• An ultrasonic deburring tool expanded the cell beyond basic handling into process automation.

Contribution

Owned design, robot programming, controls integration, vision integration, EOAT coordination, commissioning, troubleshooting, customer handoff, and training.

Results

• Switched between assembly and deburring modes in ~20 s via automatic tool change — two processes on one robot platform.
• Three vision cameras located parts fast enough to hold the 20 s assembly takt; a fourth handled feature location for deburring.
• Regrip station and process-specific EOAT kept the flexible cell recoverable and operator-friendly across both modes.
• Both modes are in real production use — assembly runs between deburring batches, which is the test the utilization case had to pass once the purchase was no longer in question.

The tradeoff

Every mode a flexible cell gains is another state it can get stuck in. Two processes, two tool sets, two gripper types and four cameras multiply quickly into recovery cases nobody wants to debug at 2 a.m. Keeping the mode logic in the PLC and the changeover fully automatic was what stopped flexibility from turning into fragility.

The failure mode I was designing against was not technical. A cell justified on utilization that turns out to be awkward to switch gets used for its primary process only, and the secondary one quietly disappears — leaving the customer's engineers holding a business case that did not materialize. Making changeover automatic and unremarkable was how that case stayed true after the purchase order was signed.

System Diagram

Sanitized architecture — a generic representation of the two-mode flow, with no customer data.