Within every manufacturing process it is very important for smooth production flow and high yield at equally high quality level to detect any faulty parts as soon as possible.  Only then it is possible to avoid rework in the production process and resulting loss of production time and thus keep productivity high.

Within the automotive industry one of the typical manufacturing processes is the assembly of complex cast parts, e.g. turbocharger, engine block or brake bodies. It is not unusual for cast parts delivered from a foundry directly to the factory floor to include different kinds of defects in the casting, e.g. blowholes, bumps, excess of material.

Today the detection of any major defects on the cast part is typically left to a random visual inspection. This visual inspection obviously increases the risk of not finding all defects on the cast part. This again increases the probability that, further down the assembly process, the individual components need to be disassembled in a time consuming process, thus reducing the productivity of the manufacturing process.

3D Quality of cast part

The goal of a joint project between the machine vision companies Photonfocus AG from Switzerland, Aqsense SL from Spain and ACT Smartware GmbH, Germany was to deliver an automatic inspection system capable of checking all incoming parts of one complex cast part at the start of the assembly process. The complete inspection system consists of 3D cameras from Photonfocus and the easy to use operating software 3D Vision Factory from ACT Smartware, which relies on the specialized 3D software libraries from Aqsense.

The new inspection system has already been able to cut down on disassembly times related to defects in the cast parts and has thus been able to increase productivity time.

The flexible 3D inspection application designed by ACT Smartware allows a fast system setup, which can be installed on an existing production line which moves the casting part on a conveyor belt or linear stage underneath a system of 3D cameras from Photonfocus and a laser. In order to minimize occlusions (shadows cast by the part upon itself) the system uses two 3D cameras.

The 3D software tools delivered by Aqsense enable a metric calibration and merger of images from both 3D cameras into a single calibrated 3D image. In the next step, the 3D Vision Factory software accesses the 3D matching library in order to compare the 3D scan of the cast part with a reference model stored on the computer. Analyzing the resulting 3D image for deviations enables the system to make a decision on whether the scanned cast part is ok or not ok. A complete 3D scan, analysis and decision making can be carried out in around a second. This enables the inspection system to be used for a 100% 3D inline inspection of complex castings, rather than only relying on an offline statistical quality control. The decision taken by the 3D Vision Factory is finally transmitted to an existing control system (e.g. a PLC) in order to physically move on or stop the cast part on the conveyor belt or linear stage.

The precision of the inspection system obviously depends on all the vision and mechanical components used, but can be as precise as a few micrometers.