INDUSTRIALIZATION SUCCESFUL USE CASE FOR VISUAL INSPECTION OF MIG/MAG WELDING PROCCESSES

The deployment in a reliable and profitable way of different artificial vision technologies that showed up on the market at beginning of 2000 turned out to be one of the main DGH's priorities to deliver more flexible automatization manufacturing process to its customers. Overall, the use of all different artificial vision technologies could be splitted into two main topics:

On one hand, the combination of robotics and 2D or 3D artificial vision let improve the performance of robotic operations by adapting movements and trajectories under real and changing conditions of production processes. Vision system will process obtained images from the environment to update robots' trajectories to meet correctly with assigned tasks. This expertise allows DGH the design of concepts for automated processes under few mechanical complexities and restrictions.

On the other hand, thanks to the analysis of images obtained on manufactured products by 2D or 3D cameras additionally with appropriate software tools, we arrange a proper deployment of quality inspection checkpoints throughout complete manufacturing process of different industrial sectors. In this last topic, introduction of AI has been a breakthrough from traditional strategies based on comparing images under finite number of measurement or appearance patterns to the possibility of teaching the vision system to make decisions by itself with learning rules that do not need to cover all infinite real situations that can occur in a normal production process (Deep Learning technology)

The use of this combination of artificial vision and AI is already widespread among different industrial sectors, but in different conversations that DGH has had with its customers and end users, it has been possible to realize that most of existing solutions available in the market are not very intuitive and therefore difficult to handle by users with little knowledge of computer vision and AI. In response to this challenge, DGH has developed a platform with software applications that let end users carry out all necessary steps of an implementation of a quality inspection system with AI and artificial vision. Firstly, a non-expert end user in artificial vision could teach DGH system in a simple way with reduced and easily obtainable image dataset. Secondly, DGH system can be tested before its implementation by confirming that it takes right decisions, and finally, once the control is running in production, an easy-to-use interface is available for production or maintenance operators. A typical installation scheme for this type of DGH platform and required hardware shall be:

An example of successful use case of implementation of this DGH platform is the installation for visual inspection of MIG/MAG welding points on vehicle chassis that was deployed for an European OEM, consisting in more than 30 controls with 16 cameras that let 100% control of welding points of all manufactured vehicles.