How Alupress cut visual inspection cycle times by 50%

Alupress wanted to automate visual inspection processes on complex aluminum parts. As the process was heavily relying on human visual inspection, cycle times were long and error rate was high.

Covision leveraged its ability to handle complex aluminium parts with its 3D scan and machine learning technology. In this process, a laser scanner is scanning the aluminum piece to generate a 3D point cloud of the piece. Then that point-cloud is compared to the perfect CAD model of the piece and defects are automatically detected.

The solution was implemented in collaboration with the automation and robotics team at Alupress.

The following defects have the highest priority for Alupress: Scratches, holes, burrs, breakouts, bulgings, indentations, cold flows

The first implementation that focused on one plant in Italy and on a set of specific aluminium parts. The goals in this phase was to leverage Covision’s 3D laser capturing technology and software pipeline and test it for accuracy and cylce times.

Furthermore the software’s transfer learning capability was assessed that allows Alupress to quickly scale learnings from one aluminum part to other parts.

Since the hardware and software setup was judged by Alupress to be reliable and accurate, we went ahead to install the vision system and software in production lines with the potential to roll it out globally.