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.
FAQ
How does DaoAI help a Tier -1 supplier improve the detection effect of glue-sealing?
DaoAI deploys self-developed 3D cameras to move with the glue-applying robot, reconstructs the 3D profile of the glue strip point by point, and calculates parameters like glue width in real-time. The vision system sends back deviations, and the decision-making unit drives the robot for real-time correction, achieving 100% online detection.
What problems did the original glue-detection method of the Tier -1 supplier have?
Previously, it relied on manual sampling inspection and visual confirmation. Problems like broken glue and missed coating could only be found afterwards. Once they entered the final assembly, the rework cost and the risk of water-leakage complaints increased significantly. Slight position deviations of workpieces could also cause the glue-applying robot to apply glue inaccurately.
What results have been achieved after the launch of DaoAI's glue-sealing detection system?
The downstream rework related to the sealant on the production line has significantly decreased, water-leakage complaints are approaching zero, and the quality data can be traced for each piece, providing first-hand evidence for subsequent process optimization.