Is the combination of AI and Vision the next big thing when it comes to robotics and automation?

Though AI is often understood differently, AI is a trending topic in automation. People know about it, people want to talk about it and people want to learn more about AI in general. Especially when it comes to the use of AI to create more flexible and versatile systems to optimize production.
Systems in general should be easy to implement and should be able to be upgraded when it comes to changes in the production environment. We would like to share a summary of an A3 panel discussion with different experts of the industry about vision technology in combination with AI and robotics. 


State of the Applications
The experts agreed that sensor level integration with AI is still very challenging and “feels” to them like a long time to get there. Although certain applications in combination with robots, vision and AI are already able to create enormous benefits.
Especially when it comes to robotic-assisted vision (eg. inspection) or when vision is guiding robotic.

An example from one of the participants (although due to confidentiality not too specific...) describes a very large, manual process in a typical industrial environment with “a bunch of people banging on things…” so the expert and with a lot of flexibility in production required.
They used three large robots to automate tasks and processes that were previously worked on manually. The integration of the robots had to be flexible to match the production requirements that were different from meter to meter.
The best position of the robots needed to be calculated and then tracked externally via
3D measurement and autonomous guidance.
In the end, a large scale process could be operated with the highest accuracy.

Another example considers a tier-one subcontractor where they had to deal with iterations and the goal to improve quality. Due to fluctuation, retirements and a lack of availability of production engineers, the challenge was to create a system that could be operated by anybody via vision programming without any internal expertise.
Is it possible to develop such a technology, where inspection plans could be created by a non-expert? Yes, with AI it was possible, so the expert.

One part of the challenge was to create a user-friendly frontend and then to identify the inspection path. The system had to be trained on what the ideal result is and then AI kicked in to compare and to improve over time.


Are we there yet?
There is progress...
We learn to solve more problems while we need to innovate and respond.
We are getting closer to the development of technology. The real complexity comes in when adding vision with sensors.

According to the experts, everyone is agreeing that automation has to be a big part for manufacturing to move forward.


What are Customers asking asking for?
Customers want speed and velocity in production. They are always comparing robotic solutions to manual handling solutions. This is only one thing according to the experts.
Another point of view for customers is uptime, which is a real benefit compared with human work.

Logistics and e-commerce have further tremendous potential for the use of robotics with vision and sensors. Then a combination of AI and autonomy is another trend in logistics to reduce risk, increase accuracy and push the limits, so the experts.

Customers can find increased functionality and solutions for all kinds of automation. Solutions, which can get up and running very quickly to perform in different production environments with the highest accuracy and speed.


What’s next?
There is an increasing ability to perform tasks people cannot perform, e.g. special welding applications.
Then there will be further integration of robotics with more ”built-in” intelligence to operate faster, eg. to perform 3D special corrections instantly.

Then there will be AI impact beyond detection. An example is to put things in intuitive
autoclassification of a defect which is then adopted by the system and automatically improved over time.

Further areas for the “what’s next” are:
- Fully autonomous automation
- Digital twin
- big data analytics
- MBE - model based enterprise
- QIF - quality information framework
- Embedded inspection data in the 3d model
- Realtime inspection feedback

All topics are in a not so distant future according to the experts…

#starttoautomate - Contact us for further information on Automation with Robot Technology and Co-Bots.