Combining RaaS, Saas and Additive to Make Robotics a Reality

3D-printed grippers, AI analytics and cloud connectivity break down barriers for robotics in manufacturing

3D printed gripper places gutter cover into box. (Source: Rapid Robotics)

3D printed gripper places gutter cover into box. (Source: Rapid Robotics)

San Francisco-based Rapid Robotics is a Robotics-as-a-Service (RaaS) provider that’s working to make robotics deployment easier with its 3D printed grippers/end effectors for the robots it rents to clients.  

Within the robotics world, end-of-arm tooling is one of the most expensive and mechanically challenging aspects. As grippers are generally the part of the robot that interacts with products, issues with grippers and end effectors can introduce risks to the production process. For this reason, some companies specialize in end-of-arm tooling, selling grippers that cost tens of thousands of dollars. 

In keeping with the RaaS model that aims to simplify and reduce the costs associated with robotics adoption for companies, Rapid Robotics has taken a novel approach to grippers. According to Rapid founder and CEO Jordan Kretchmer, the company’s goal is to provide inexpensive, accessible robotics solutions that are highly reliable, easy to deploy and deliver positive results for clients. The company’s RaaS services offers more than just the robots themselves, but also an entire suite of software, cloud-based services, computer vision and sensors that work in conjunction to keep rental prices between $4 and $12 per hour—a fraction of the cost of human labor. 

Kretchmer said that end-of-arm tooling customization was a primary focus for the company from the beginning because of how essential it is to clients’ maximizing throughput. Yet, because end-of-arm tooling requires specialized tools for every application, the added costs pose a barrier to the RaaS vision. With Rapid wanting to build on its promise of fast and low-cost robotics deployment, the team turned to 3D printing to create flexible, inexpensive grippers that can work for virtually any process. Instead of companies spending around $150,000 on the grippers they need for their factories, Rapid can 3D print grippers that cost $75 each, and take only 12 hours to produce.  

“I think the primary thing is flexibility,” said Kretchmer. “Our system can very quickly print new kinds of tooling. About 98 percent of U.S. manufacturers consider themselves high-mix facilities, meaning they’re trying to get maximum machine utilization. But the problem right now is you don’t see robots in those facilities, and that’s because of the lack of flexibility.” 

3D-printed grippers could be game-changing for manufacturers that either haven’t deployed robots or use models that are bolted to the floor and restricted in application. Rapid’s self-aligning, 3D-printed customized grippers can be quickly and easily swapped out by clients, which could result in major increases in productivity. 

According to Kretchmer, Rapid offers both flexibility and reliability. He said the design process for the 3D printed grippers is data-driven and results in a perfect fit for machines, 100 percent of the time. Rapid’s engineers manage a library of over 100 generic grippers that can be used as templates for new customized designs.  

“Instead of spending two weeks designing a gripper like we did in the beginning, most of the ones we design now just involve tweaking the more ubiquitous designs. So in a few hours we can have a new gripper. The larger our library gets, the stronger that IP becomes for us and the faster we can do more complex deployments.” 

This contrasts with traditional gripper fabrication, which can leave clients waiting months for a new gripper that sometimes needs to be shipped from overseas. 3D printing allows Rapid to get clients fitted with new grippers within hours.  

“It’s going to have a huge impact on the market,” said Kretchmer. “Because it’s the lack of flexibility that’s the primary reason why manufacturers don’t utilize robots. We’ve demonstrated that our grippers are highly reliable—we’ve never had a gripper break on a customer.” 

Rapid currently 3D prints grippers in San Francisco and Detroit. However, clients who have sufficiently advanced 3D printers can also print their own grippers for use with Rapid’s rental robots.  

Another advantage of 3D printed grippers for robots is that they’re lightweight. Heavier traditional grippers cut into a robot’s maximum payload, increases the risk of breakage and can drive up energy costs. Rapid had initially printed grippers as solid structures, but has since evolved to more sophisticated designs with complex internal structures that are rigid but also very lightweight. Kretchmer estimates that some of the 3D printed grippers weigh four times less than one produced by traditional manufacturing methods. 

The applications of 3D printed grippers are virtually limitless because the technology allows for geometrically intricate designs that can’t be achieved with traditional manufacturing. This means that clients can deploy robots for more complicated tasks. For example, one of Rapid’s clients used robots to produce AeroPress coffeemaker parts. One of the tasks involve stamping the AeroPress logo onto a specific side of a cylindrical part. Rapid’s AI and computer vision ensures that the gripper is oriented correctly when grasping the base of the hexagonal-shaped part. Although the vision system can only be accurate to a certain degree of precision, the fact that the gripper itself matches the exact angles of the hexagonal shape, means that it successfully grasps the part even if the vision were off by two millimeters—making the accuracy effectively perfect.  

Another client produces gutter covers and previously had 12 people operating machines daily. However, Rapid created an inexpensive end-of-arm tool that can accurately grip tiny holes in the covers, saving the client significant overhead costs. 

3D-printed circuit board handler. (Source: Rapid Robotics)

3D-printed circuit board handler. (Source: Rapid Robotics)

In another example, a client produced circuit boards that involved a machine placing sheets of copper in 14 different sizes. The challenge was that the gripper had to be able to grip the different sized sheets without knowing which dimensions would be next on the production line. Additionally, having to swap out the grippers to accommodate the varying sized sheets would derail output. So Rapid harnessed its cloud and AI technologies to update and activate the robotic arm’s motion and adjust the suction cup grippers’ placement exactly in the right place to accommodate the varying sheet sizes.  

Robotics-as-a-Service vs. Robots-as-a-Service 

According to Kretchmer, RaaS originally stood for Robots-as-a-Service, meaning that providers would lease robots to clients for a monthly fee. However, the service didn’t necessarily come with any other necessary technologies like speciality end-of-arm tools, AI and vision tools or cloud services—just the robot.  

“We’re able to provide something far beyond just the robots. It’s not just the arm itself but all of the other additional components that work seamlessly together. It’s the technology stack, the IP and the intelligence that’s connected to the system that makes it far more valuable,” he says. 

All of Rapid’s robotic systems are based in the cloud. That means that the company remotely performs maintenance updates and respond to changes on the factory floor. The AI models can be updated remotely so that a clients’ robotic arm can execute on new parts or new positions quickly. It also enables Rapid to provide 24/7 support and examine every single robotic action and chart performance over time. If a customer submits a support ticket for a bot, Rapid can immediately examine the performance history and pinpoint where things went off track. Additionally, Rapid’s cameras allow the support team to see the cause of the problem. For example, if an arm stopped moving at 10:02 am, Kretchmer said it would have probably been caused by a person breaching a robot’s safety boundary inadvertently. Rapid’s support team can  reset the systems, and within 30 minutes the robot can be back up and running versus other RaaS companies that would have to conduct an onsite visit.  

“Any of our customers will say that Rapid is an extension of their team,” said Kretchmer. “They don’t need to know anything about robotics. They don’t need to program the robots. We’ve removed all potential points of failure and removed all potential points of human error.” 

Kretchmer anticipated that competitors will try an emulate Rapid’s RaaS framework. However, he said that one of the company’s market advantages is that it started out as a software business and has cloud architecture and AI capabilities that other companies with a more hardware focus don’t have.