Edge AI Solutions for Industrial Assembly with Atlas Copco

Today, a high level of expertise is required to set up the system correctly, troubleshoot, and optimize the production process. This complexity makes it very hard for a non-expert to adjust the system in the best way to the process, identify and solve errors very quickly, or schedule maintenance accordingly. The system’s behavior is also highly dependent on local environmental conditions such as adhesive properties, type of application, and robot programming, which prevents us from using pre-trained AI models without adapting them to local needs.

The system should hide this complexity from the customer by making intelligent decisions based on customer-specific data. This includes but is not limited to automatic root cause analysis in case of failures, constant wear monitoring for predictive maintenance, improvement of production quality, and waste reduction. The ultimate vision is that only a start button needs to be pressed; everything else is done automatically and requires as little user attention as possible. An ideal solution should also work with limited computational resources and without a constant internet connection.

Atlas Copco is looking for edge AI partners who can enable self-supervised learning in cases where customer-specific data is considered sensitive and cannot be processed in a server. We are open to various solutions, including software and hardware innovations. Though an ideal solution solves the end-to-end problem, we are also available for solutions to solve more specific sub-problems.

In the pilot phase, the solution could be applied to a few test facilities but later scaled to a broader use in Industrial Assembly Solutions’ products. Moreover, Atlas Copco can provide access to various other industrial assembly opportunities and major industrial OEMs.