Improving quality assurance with machine vision and Industrial grade AI

AI is the hot topic of today in public discussions and inside the board rooms. Companies wish to integrate AI in their operations but finding ways that actually streamline processes and increase efficiency might be difficult.

We at Vaisto have been working for years to help industrial companies make their operations smarter. Integrating AI solutions to existing processes and production lines with smooth and agile project management is our core expertise.

For many industrial companies, automated quality inspection with machine vision has the potential to boost quality and efficiency of manufacturing processes. The complexity of the quality inspection processes and integrating machine vision to them can make business calculations and project planning tricky at times. We know how to approach these challenges and split the projects down to manageable steps.

Let’s take an example of a case, where the customer wants to improve quality assurance and inspection on an industrial production line by applying machine vision and AI. The project can be divided into six steps.

1. Use-case analysis – Elaborating on the case with customer and specialists

Our machine vision solutions are built on deep understanding of the specific customer production characteristics. Before we start a project, there’s a bunch of what and why questions to be asked.

– What are we inspecting and why? What are the precise criteria for success?
– What are the exact defects to be detected? What are the tolerance levels for those defects?
– Which environmental factors should be taken into consideration?
– What is the desired speed and accuracy of the inspection?

Together with the customer and our specialists, we build a common understanding of the case problem and the desired outcome.

2. Identifying the right hardware to use – Technology, solution blue print

Once the machine vision potential is validated, the optimal technology and architecture for the specific use case is chosen by our technical experts and customer team. The industrial environment and specifics of the project set a frame for choosing the right cameras, sensors, and lighting. Equally important is the choice of computing hardware to ensure that data from the cameras and sensors can be processed at high speed and the algorithms run smoothly.

3. Feasibility study – business case and technology validation

To assess the business case of our solution, we collect representative sample data from the customer’s production line to test the suitability and reliability of the chosen technology for real-life conditions. Based on the data, a business case can be either justified or not by analysing the frequency and impact of defects, the potential for error reduction, and the increase in efficiency through automation. The initial data and technology analysis also allows us to identify the best way to initiate the project to demonstrate the value and viability of the machine vision solution.

4. Preparing the project with customer

Before starting the deployment, careful preparation for the project is done together with the customer. Any potential challenges are addressed and the integration with existing production systems is confirmed. We want to make sure the implementation goes smoothly, so clear milestones, roles and responsibilities are defined, and it’s made sure that both parties agree and are aligned on the success metrics and what constitutes a successful outcome.

5. Deployment in an agile way

We apply agile delivery methods for solution deployment. The agile model allows the deployment in small manageable iterations, providing tangible results quickly and frequently. The goal is to get the machine vision system operational in a limited capacity as soon as possible, demonstrating its value and receiving results in early steps to make continuous feedback and quick adjustments possible.

6. Full deployment and piloting side-by-side with customer

Once the initial deployments have demonstrated value, the solution is deployed to production. This involves integrating the machine vision system seamlessly into the existing production line, ensuring minimal disruption. The system is piloted side-by-side with human operators, to allow comparison to human inspection and offering real-world feedback for continuous improvement.

By following these steps, we can implement machine vision and AI solutions to quality assurance processes in industrial operations to provide business value and competitive edge to our customers. 

Want to know more?

Contact us