Physical AI is poised for widespread adoption in some applications: CreateMe CEO

Applications of AI


This voice is automatically generated. Please let us know if you have any feedback.

CreateMe, an automated soft materials manufacturing company, has created a system that uses physical artificial intelligence to revolutionize the way textiles are manufactured. To do this, the company is replacing traditional sewing, which is particularly difficult to automate, with digital bonded structures that leverage robotics, proprietary adhesives, and an AI-driven manufacturing system.

The company says its modular engineering robot assembly system is the world’s first autonomous tailoring platform that combines hardware and software, “accelerated by physical AI to deliver dynamic variability and flawless precision.” However, challenges remain, especially when it comes to handling cloth, which is not stiff and difficult for robots to handle consistently.

CreateMe CEO Cam Myers has been awarded 25 patents for apparel automation technology developed by his company. Prior to founding the company, he was part of the founding management team of Group Commerce, a venture-backed e-commerce platform that was eventually acquired by Blackhawk Network, and previously held positions at DoubleClick and Allen & Co. He holds bachelor’s and master’s degrees from the University of Cambridge and an MBA from Northwestern University.

In a recent interview with Manufacturing Dive, Myers spoke about the current state of physical AI, how some companies are starting to integrate the technology into their manufacturing processes, and the steps other companies can take to see if it works for them.

This interview has been edited for clarity and brevity.

Manufacturing Dive: How is your company currently using physics AI? What are your plans for the future?

Cam Meyers: We’re using some kind of basic physical AI model, but we have a very unique perspective because we’re focused on deformable materials. Like this humanoid, you don’t have to move horizontally, like washing the dishes, then folding the laundry, then taking out the garbage. We need to be very good within narrower constraints. How to handle the fabric [including] Type of fabric.

We use some basic models, but we specialize in specific use cases…I think this is where rubber comes into play in manufacturing as a narrower application of physical AI…so it’s not just point cloud to point cloud positioning of mobile robots. It can still be non-deterministic, but within a range of more constrained outcomes.

Following this principle, we collect data in three ways. We have a typical remote operation. Our molds are semi-automated using human operators; humans do the work. Place the fabric in the mold and fold it to create what is essentially a neckband found on a crew neck. [T-shirt]. we [also] There is an actual human operator [manufacture a] The product…has captured training data, supplemented with two other data channels.

[We also] It has a data channel with remote control that mimics the same motions a human operator would make to place fabric on a device as a robot would. And we’ve developed an interesting handheld grip technology…inspired by medical applications for robotic hands.

We have this training data and at the end of the year these roadmaps will converge and at that point we can start removing steps that still have human operators.

What do you think about other manufacturers using physical AI?

In my opinion, physical AI will be more tangibly implemented within the next three years in manufacturing, where the ROI is highest. What I’m really saying is that we’re going to see more adoption in narrower applications in the short term.

I think some of these more general purpose humanoid robots are a little further down the line. of [best] Opportunities for physical AI applications compared to other types of more traditional robotics and automation in the industry include: [physical AI] It can still start to become a little non-deterministic, but it’s not completely unlimited.

These tools must be selected and placed. But the next evolution will see a significant increase in variability. For example, a warehouse. It’s just a matter of finding the right springboard.



Source link