Sealing is the first line of defense for body tightness and NVH; one invisible break can surface months later as a water-leak complaint. With DaoAI 3D robotic vision, a Tier-1 supplier moved bead inspection from sampling to 100% inline and let the robot correct its path on the fly.
Sealant quality directly determines a vehicle's water resistance, dust sealing and NVH performance, yet bead defects are often invisible to the naked eye. This Tier-1 supplier's door-panel sealing line previously relied on manual sampling and end-of-cycle visual checks, so breaks, skips and uneven bead width could only be caught after the fact. Once such parts reached final assembly, rework costs and water-leak complaints multiplied. Minor incoming-part position deviations also caused the fixed-trajectory dispensing robot to lay the bead off-target.
Brain-Eye-Body Loop: Inspect While Correcting
DaoAI mounted its in-house 3D camera to track the dispensing robot, reconstructing the bead's 3D profile point by point and computing width, height and continuity in real time. Vision acts as the eye, feeding deviations back to the decision unit that serves as the brain, which then drives the robot as the body to correct its path on the very next segment, forming a sub-millimeter hand-eye coordination loop. Inspection and dispensing complete simultaneously, adding no cycle time.
- 100% inline inspection: every part, every bead segment fully checked, eliminating sampling blind spots
- Every break and skip caught: continuity gaps, start/stop overflow and voids trigger instant alarms with coordinates
- ±0.5mm bead-width accuracy: out-of-spec width/height auto-judged with traceable data
- Real-time path correction: robot trajectory dynamically adjusted to incoming pose and bead deviation
Inspection is no longer a gate after dispensing; it is a closed loop that happens at the same moment as dispensing.
After deployment, downstream rework related to sealant on this line dropped sharply, water-leak complaints approached zero, and quality data became traceable part by part, providing first-hand evidence for ongoing process optimization.