AI and Robots: Not what you think
Brilliant new blog by Jim Lawton for Universal Robots: AI and Robots: Not what you think – Enjoy the read!
There’s a lot of excitement about what artificial intelligence (AI) can do in manufacturing. Depending on what you read – and choose to believe about what you read – AI-driven robots are able to autonomously make decisions about what work gets done, how it gets done and who does it – or there are decades of work yet to be done before we see a material impact.
Personally, I think we’re somewhere in the middle, as manufacturers, pragmatists that they are, design and implement manufacturing strategies in a very deliberate way to achieve business requirements, focusing ongoing efforts to make key processes better and better. I think that collaborative robots (cobots) will play a larger and larger role in accelerating that progress. The AI that cobots possess makes them so much more than just machines for dirty, dull and dangerous work.
So let the world watch and wait for artificial intelligence that will enable wholesale change in how we drive, care for our aged, teach our children and more. Manufacturers don’t have to wait for artificial intelligence-driven robots to help them make their operations better. It can – and is – happening right now.
Improve, Improve, Improve:
In manufacturing, continuous process improvement is part of an operation’s DNA. Whatever is being done well today, can be done better tomorrow. In these environments, AI-driven robots can make meaningful contributions to process improvement from the day they are deployed.
I’ve written a lot about how cobots are much easier to use than traditional automation solutions based on the simple fact that they are safe enough to work in close proximity to people and don’t require hours of integration and custom programming in order to work on a task. There are practical applications of intelligence that make this possible, including these:
This ability makes it possible for an AI-driven cobot to, for example, detect changing workspace conditions and to monitor and optimize its operation. Key performance indicators (KPIs) are essential for monitoring your robot and for guiding decisions that will help you get the most bang for your cobot buck, year after year. Robotiq Insights is an example of software providing real-time data to monitor, troubleshoot and improve production.
With this ability, an AI-driven cobot can recognize the presence and orientation of parts; perform inspection and dynamic pick and place tasks as well as read results from testing equipment and make decisions accordingly. The 3D-PickIT camera is one a growing number of AI-powered vision solutions, enabling cobots to pick up randomly oriented parts of all shapes and sizes.
Using this trait, an AI-driven cobot can adjust task orientation as machines move; adjust the force control required to pick parts from a stack; detect and evade collisions and respond to errors with retry strategies.
An AI-driven cobot that can predict and diagnose failure conditions; identify patterns in ongoing operations and apply insights gained to drive better performance. Unlike traditional factory robots, RightPick from RightHand Robotics handles tens of thousands of different items using a machine learning backend coupled with an intelligent gripper that works in concert with the UR cobots.
An AI-driven cobot can be put to work in hours; it is also able to re-use task information, share and communicate it with other robots and motion control systems.
AI-driven robot can control other machines; orchestrate the activities and improvements in nearby equipment.
The Haas CNC Integration Kit from VersaBuilt enables Universal Robots to easily execute any machining program stored on Haas CNC while maintaining all Haas safety interlock features.
What matters most to manufacturers is how these translate into operational improvements. That’s easy. With onboard intelligence, cobots can realise when something is not working and will stop before damage is done identify ways to improve the way a task is done collect data and perform analytics to help users make decisions around process improvement.