Digital Twinning for Common Robot Applications
Posted: 10/31/2022 7:47:53 PM by Sarah Mellish
Topics: Arc Welding, Handling, New Technologies, Software
From part fabrication, assembly and testing, to palletizing and more, the use of digital twin technology for robotic control is optimizing overall performance for advanced robotic applications. Generally speaking, a digital twin for robotics is taking an accurate 3D model of a robotics operation into the virtual world, where it can be tested and enhanced before implementation into an existing workcell. Whether gained via computer simulation, preprogrammed algorithms or iterative machine learning, digital twin data can help make predictions about future outcomes, yielding the utmost accuracy, safety and efficiency for a specific task.
Material Handling
Associated with a significant uptick in factory and warehouse productivity, the use of digital twin technology for workcell design and optimization is extremely impactful for end-to-end automation. From physics modeling of the process equipment to that of the materials being processed and other components, robust technology with dynamic software can help account for multidirectional requirements. Plus, challenges for throughput quotas, positioning accuracy and flexible response to outside conditions can be met head-on via mathematical modeling and algorithms.
“Looking ahead, robotic technology will be improved to better align with digital twins,” says Chris Caldwell, Product Manager at Yaskawa Motoman. “These anticipated improvements will significantly improve capability with artificial intelligence (AI) and machine learning software, as well as offline programming and optimization efforts.”
As more companies embrace computational algorithms into their digital twin experience, a greater number of workflow and operational challenges should be solved pre-emptively with greater efficiency. This is especially true for complex and/or highly repetitive design processes.
Robotic Welding
“For robotic welding, the digital twin concept can be as simple as using offline programming software to find the ideal programming path,” offers Josh Leath, Sr. Product Manager of Welding. Whether trying to reduce air-cut time, minimize cycle time, produce high quality welds or something else, rendering a digital model that can be overlayed into an augmented or virtual reality – where high-value decisions about design, intent and instructions are made and validated before being used in a tangible production environment – can be extremely beneficial. While finding the “perfect” process is not automatic, using intuitive platforms and methods, where the computer can be “taught” over time with some sort of feedback (i.e., human, sensor technology or post-process scanning), can ultimately enhance the robotic application.
Keep in mind that production parts are never the same as the model, so it is important for manufacturers to use appropriate technologies, as needed, to adjust for the difference between the digital world and reality. Innovative sensor technology, or even 2D and 3D cameras, can help fine-tune minor fluctuations. However, if a part is dramatically different from the model, or an operator has no idea what the part should look like, an advanced digital twin solution can input the real-world part through scanning technology (i.e., 3D profile sensors, cameras, etc.) and create a model for the digital simulation. Note: this usually requires an extremely powerful computer to process the hundreds/thousands of simulations needed to achieve the best solution. However, even this requires the computer to be trained through feedback.
Where feedback is concerned, closed-loop setups allow certain sensors to instantly provide training feedback for another output response. A common example of this would be with vision or LiDAR systems in a self-driving car. The car vision can detect road markings and “see” the road, telling the steering components how to immediately adjust, when needed. Keep in mind, tire pressure, slick rocks, loose pavement, etc. could impact the accuracy of the response, so the car PC needs to understand how to compensate through that closed-loop feedback.
The same goes for robotic welding. If a human can see a part, then create a model of that part, but is still producing a bad weld in real life after a “successful” simulation, additional variables may need to be detected. From material type to rusty metal, different types of sensors may aid application success.
Moving Forward in the Digital World
Manufacturers that successfully incorporate Industry 4.0 technologies, including the digital twin concept, into their operations can enhance operational productivity, human safety, product quality and more. This data-driven approach is especially helpful for minimizing the robot programming learning curve – ultimately accelerating robot uptime and product throughput.
If you think the use of digital twin technology can unlock new capabilities for your manufacturing operations, or if you have questions about upgrading technology on your production floor, reach out to our industry experts today.
Sarah Mellish is a Marketing Content Specialist
Connect on: