September 29, 2021

GrayMatter400x275GrayMatter Robotics, which develops software and artificial intelligence algorithms to create smarter robotic assistants, has announced it raised $4.1 million in seed round funding. The round was co-led by Stage Venture Partners and Calibrate Ventures, with participation from 3M Ventures, OCA Ventures, Pathbreaker Ventures, and B Capital Group.

Founded in 2020 by Ariyan Kabir (CEO), Brual Shah (CTO) and Satyandra (S.K.) Gupta (Chief Scientist), GrayMatter Robotics develops proprietary AI algorithms that enable industrial robots to program themselves. The company packages its software with commercially available robots, sensors and end-of-arm tools to deliver smart robotic assistants to manufacturers for surface-finishing applications, such as sanding, polishing, deburring, and spraying. The robot system can learn to work on a new part it has never encountered before in a matter of minutes through the use of machine vision and path planning. Users can roll in a new part and the system will scan it to identify its geometry, pick up the right tool to process the part, access a database to figure out its material and which process to use, and then program the robot arm in the correct motion and execute the operation.

“It is shocking that millions of people still manually work on extremely tedious tasks that can be harmful to their health; younger generations do not want such jobs,” said Kabir. “We want to improve shop workers’ lives, enhance their productivity, and enable them to focus on higher-value tasks. Manufacturing drives our economy, and without automating surface finishing and treatment, there is a significant risk the global economy may suffer due to an increasing labor shortage.”

GrayMatter Team800px

Members of the GrayMatter Robotics team.

Most of the large industrial robots being used can only handle specific parts that are the same every time - any variation in the part’s design or even production errors can throw off the robot and require it to be reprogrammed. This leaves surface finishing tasks in the hands of human workers, due to the variations of parts that need to be sanded, sprayed, or deburred. GrayMatter estimates that more than 1.5 million workers in the U.S. are still doing these tasks manually.

“Finding and retaining employees willing to do sanding has always been challenging, and in the current labor shortage it has prevented us from growing,” said Francis Hu, president of Performance Composites, a GrayMatter customer. “GrayMatter’s Scan&Sand automation solution has allowed us to free up current sanding employees to work on higher-value tasks and provides a good return on investment.”

With the seed round funding, GrayMatter said it will scale up hiring, accelerate development of next-generation products, and engage with a wider customer base.

Exclusive Q&A with GrayMatter CEO

Robotics-World recently spoke exclusively with GrayMatter Robotics CEO Ariyan Kabir and Kevin Dunlap, a partner at Calibrate Ventures, about the seed round funding.

Robotics-World: Congratulations on the funding – can you give us a sense of the background of the company and what led you to figuring out how to automate the surface finishing tasks in manufacturing?

AriyanKabir GrayMatter150pxKabir: The co-founders of the company, we all met at the University of Maryland, where we were doing our Ph.Ds. We were working with Professor Satyandra Gupta, who is also a co-founder of GrayMatter. In 2016, Gupta moved to [the University of Southern California] to direct the Center for Advanced Manufacturing at USC. It was a great opportunity for me and Brual to also move to USC and be part of the center. That gave us a unique opportunity to work very closely with the industry. Every week, we had visitors coming in from small and medium-sized businesses, as well as large enterprises like Lockheed Martin, Northrop Grumman and SpaceX.

Initially we were developing some really cool algorithms, making robots do many different things, but what truly opened our eyes was when a person from Lockheed Martin told us how painful it is to manually make the 40-feet long Sikorsky helicopter blade. We were totally surprised because every day we saw on YouTube how people were using robots, like Toyota or Tesla using hundreds of robots to make cars. It was really surprising to learn that someone was waking up at 4 a.m. and then spending 10 to 12 hours every day of their life doing nothing but sanding and finishing operations, sometimes sweating in 110 degrees. We also learned that between two and five years, people would get some health injuries like carpal tunnel, shoulder or back injuries.

That was truly eye-opening for us. We started focusing our attention into combining our AI technology with existing mature robots, sensors and tools to create this turnkey solution that the end user can use like an appliance, where you just give a part to the robot, it scans the part, understands the geometry, auto-programs itself and starts execution. The sensor fusion process is constantly taking feedback from sensors to make online adaptations and adjustments during the process to ensure quality, consistency, and make sure it’s a safe operation.

RW: What is it about surface finishing that makes traditional robotics so difficult to apply to the problem?

Kabir: For most of manufacturing, it's a high mix, with high variability and variations coming from the geometry and material distribution, and the manufacturers are working on multiple SKUs in the same week, sometimes even on the same day. As a result, they can’t really afford to spend the capital or the time to program a robot every single time to work on a part.

For example, one of our customers who is making composite parts, if you closely inspect the parts, you’ll see a -1 cm of deviation in the geometry from part to part, and the traditional robotic approach does not really scale. The way we approach the problem is to enable the robot to understand those variations for every single part, and program itself so that the end user can have someone running the system without any knowledge of robotics or engineering or automation.

RW: Does the size of the part matter for the system – is this mainly for large parts, or do you have a size range that the system falls within?

Kabir: We can capture a wide range of part geometries and dimensions, but typically most of the attraction has been on large parts – things like aerospace parts, or something on vehicles, buses, trains, trucks, and marine equipment. 

RW: After the sanding or other process is completed, would a customer then perform touch-up or inspection on the part, or is it completed by the robot alone?

Kabir: In our applications, we enable the robot to do an inspection in addition to the initial surface finishing process. The way we approach the problem is that you enable the robot to take care of about 90% of the large surface area, and then about the last 10% could be done by a human, depending on the part size, geometry and process.

 

 

RW: Is your software robot-agnostic and end effector agnostic?

Kabir: Right now we are partnered with Fanuc and Universal Robots, and are offering our solution to customers with these two robots. However, the software will support other robot vendors, including ABB, Epson, Yaskawa, and other robot vendors. For end effectors, we work with FerRobotics, PushCorp and ATI for compliant end effectors. We also work with 3M for sanders or abrasives, and LMI, Keyence, Cognex, and Hexagon for different sensors.

RW: A question for Kevin on the Calibrate Ventures side. What was it about GrayMatter’s approach that stood out to you, in terms of investing in the company?

Dunlap: For us, it was really two-fold. It was the team, and the technology is really at a point where it can solve real needs for manufacturers. It’s that perfect mix of technology and experience, especially at the early stage.

We’re also investors in a number of robotics companies, and we’re seeing the labor shortages and wage increases that are happening in lots of industries, and it’s definitely happening here. You see industries where people just aren’t showing up for work anymore, maybe they get paid more down the street, or they’ve decided they’d rather do something else with their lives, and companies can’t find labor. This application also is that dirty, dangerous and dull job that robotics can solve. So we can take these workers and upcycle them within an organization, and have them do other things that add more value within the business. We also know that material removal and advanced manufacturing is the next place that builders and manufacturers are focusing their efforts. That’s great for companies like GrayMatter, because they really want to work with us – they know that this is the next application for them to move a lot of robot arms.

This technology, five years ago, wouldn’t have worked, but the robot arms are much more advanced – even the cobots that are coming out now are much more advanced from a payload perspective, sensing, feedback, and computer vision is much cheaper and at a better scale. It feels like for us that this is the perfect combination coming together.