Analog's Mechatronics Engineer: Hardware Powering AI Inference | Researcher Conversations at GTC

Hinzugefügt: 2026-05-15

[00:00:09]Everyone, welcome back. This is Jordan Nanos here for semi analysis and cell media at GTC 2026. I'm here with Misha F Musa.

[00:00:17]We're going to have a little chat about uh yeah, what's new? Welcome. Yeah, thank you.

[00:00:22]So, you guys had a interesting demo on the show floor this this week and I know you're part of a a new division within ADI. Can you give me a just like a high-level summary of what you guys are working on? Yeah, so I'm with the Emerging Tech Hub in Analog Devices. Uh we're a relatively new group um focused on uh robotics and edge computing within the company.

[00:00:41]Uh we're a 60-year-old company uh providing semiconductor and chip level solutions. Uh however, our group works uh a little bit higher level at the system level trying to see how we can integrate our existing technologies into robotic solutions and kind of push the edge of technology there. Yeah, cool.

[00:00:57]So, take me through what are some of the concepts or ways you see robotics would be useful in these uh applications?

[00:01:03]Yeah, so uh and specifically talking about one of the demos we're showcasing uh here at GTC, um we're using uh our in-house built tactile sensing solution.

[00:01:15]Um essentially, it's a sensor that fits into like the fingertips of robot grippers or hands or whatever types of dextrous manipulation we're trying to accomplish.

[00:01:23]Um and it gives us a sense of uh the forces we're applying as we're grasping on objects. It allows us to be delicate if we need to handle um for example, the fiber connectors in a data center type application. Um and so, what's nice about the sensor as well is that it's a multimodal sensor.

[00:01:41]And so, not only are you getting force feedback and the pressure sensitivity, but it also has accelerometers um and some vibration sensors that can give you a better picture of what's happening at that interaction. Okay. So, literally plugging in cables. Yeah, that that's one of the applications. Um we work very closely with a lot of different partners in our in our team.

[00:01:59]It's kind of a co-innovation space within EDI.

[00:02:02]Um, and one of the key applications that we're noticing is trying to automate uh, some of the maintenance uh, that's done inside of data centers.

[00:02:09]>> Mhm. Um, so going in and disconnecting cables, polishing fibers, um, and all of that requires a high level of precision as well as sensitivity to these types of connectors. Mhm. Um, and so we think uh, some of the solutions we're providing there. You're going to need a robot using a one of the clickers right now to clean.

[00:02:26]Uh, at the moment, no. Uh, but that's kind of the the step where we're trying to push. We're we're also trying to engage a lot more in the robotics community. So we're developing a sense uh, a set of benchmarks that'll be physical boards that are analogous to types of problems we see in data centers, uh, automotive cable harness uh, assembly, um, gearbox assembly.

[00:02:47]Uh, it'll all get released to the public, um, open source so that researchers and people in industry can work on these problems and be able to share um, knowledge around how we go about solving these problems. But wait, yeah.

[00:03:00]So the benchmark of the leaderboard is going to be a robot in a room set up. So >> Not necessarily the robot. You're welcome to choose whichever robot you want to use to solve the problem. Okay.

[00:03:09]>> It's just a board that sits on the table and the robot can go in. You program it however you choose to, whether it be classical control or using newer AI-driven policies um, in order to solve these problems and hopefully be able to um, allow others to learn from from what you're doing as well.

[00:03:24]>> Yeah, a little bit different than grade school math benchmarks though, I think it's Yeah, a little bit. Um, and then but what's nice too about what we're doing is that not only are we providing somewhat of the the physical device itself, but also we're working with uh, key partners to develop high-fidelity simulation assets Mhm. that allow people to train their policies in simulation.

[00:03:42]Um, we'll have we'll be able to provide simulations for our sensors as well, so like our tactile sensing um, or time-of-flight solution that's able to create a a depth scene um, or a depth estimation of the scene we're looking at and all of that can be used together to create really nice robotic solutions.

[00:03:59]>> Yeah. So, what does this look like as a product that you guys would sell? Would it be, you know, a software product people can run on generic robots? You would sell a robot itself? Uh so so we're not we're not a robotics company by any means.

[00:04:12]We we provide the solutions that enable roboticists to do what they do best. We we control a very precise signal chain all the way down through our analog to digital converter, CMOS sensors, now tactile sensing, time of flight sensing, and we're really pushing um basically at the edge of what these devices can do to enable roboticists to go in and solve these problems using our sensor and our other types of technologies. Got it. Makes sense.

[00:04:38]Where do you think you guys go from here? More benchmarks, more examples?

[00:04:42]>> Yeah, I think I think robotics is a long-standing challenge in terms of being able to collect data with robots. And so being able to provide this type of solution where if we have very accurate simulation models, it minimizes what we call the sim-to-real gap.

[00:04:58]Um and then from there people can, you know, choose you're free to choose how you want to solve those problems, but we just want to be the company that's enabling people to be able to do that. And if you choose to use our sensors, we think that's probably what's going to be best.

[00:05:13]But again, it's it's all up it's all about enabling people to do what they do best with their robots. Makes sense, yeah. What are some of the applications I guess that you're most excited about.

[00:05:21]I guess the uh yeah, the cleaning of fibers is one that comes up all the time in data centers.

[00:05:26]>> Definitely, yeah.

[00:05:28]We're we're we're pretty excited about the the data center particularly. Within our group, we have some key partners that we're working with to see how our system of a group can kind of integrate some of the solutions and be able to test it out. Um a big thing for us is being able to understand the problems that people are having in the industry and what they need to solve and how we can provide the technology that will enable them to solve those problems. How do they do that? Do they provide data? Do they provide just like a spec that you have to design towards? Like >> Well, yes, and that's the kind of the unique feature about our our team. Um, it's kind of designed around this idea of being a co-innovation hub.

[00:06:05]>> Uh-huh. So, they're more than welcome to bring their engineers into our lab.

[00:06:09]Uh, we work together to try to solve the problem, whether it be they're bringing in new tech from their company um, and seeing how we can support that, or we're showing off some of the new tech we have and how we can enable them to solve their problems as well. Do you have any examples of like collaborations that you can talk about? At the moment, no.

[00:06:26]Uh, but it's but it's ongoing and we're pretty excited about uh, what we're doing. Makes sense.

[00:06:31]How about um, GTC in general, right?

[00:06:33]There's lots of work with Nvidia and GPUs and AI and training models and stuff like that. Are you guys um, doing anything unique with the uh, on the model training side, you know, in terms of like maybe how you run the system infrastructure or data center for yourself? Yeah, um, so I wouldn't necessarily say we're doing anything necessarily unique on the the training side, per se, but when it comes to actually deploying the model and doing inference, um, we're we have a team dedicated to edge computing that's trying to push the technology to the point where we can take these large models, distill it down into a form factor that allows uh, people to run it on their robots, uh, in your autonomous vehicle. Uh, yeah. Yeah, so what does that work like?

[00:07:14]The chips are getting more little more powerful every year. They're getting a little more energy efficient. Meanwhile, the models, the dis- distillation's getting a little bit better. So, they're kind of converging together. Uh, that that's the idea is that we're we're we're taking these large models, putting it into a compact form factor uh, that can sit on the edge um, in a more power efficient way.

[00:07:34]>> [clears throat] >> Um, and then that that enables you to, you know, not have to be having a, you know, super big GPU box sticking off the side of your your robot or whatever.

[00:07:43]All of it can be, you know, put into a real small form factor.

[00:07:47]Yeah. Are these Nvidia GPUs or are these custom chips?

[00:07:50]Uh Kind of varies?

[00:07:53]Yeah, I think it depends.

[00:07:55]I'm not too much involved on on that side of the this this the story when it comes to the edge computing.

[00:08:01]But Uh It can be pretty complicated just to get a model to run on some of these custom chips that people are building. So, I'm curious if there's a way in which the chip itself might be co-designed with the model that it's actually going to be running. Yeah, I think that that's a big part, too. And again, when it comes to the co-innovation that I talked about is that we're not only that in the robotic space, but also in the edge computing.

[00:08:23]So, we talked to the big players in the space and see how we can integrate near technologies or sorry, their models into the fabric of whatever we're trying to design. Yeah.

[00:08:33]>> And of course, it costs them SOCs or ASICs and all that.

[00:08:37]Right, yeah. I guess there's a lot of cases where people might be training really big models and just have no idea how they I guess it solves the task, but let's just what? Wait 3 years for the chips to Exactly. Catch up. Exactly. And so, we're we're we're thinking a little forward and like, where do we see these models getting deployed in the robotic space, autonomous cars, home [snorts] robots, yeah. Awesome.

[00:09:00]Well, yeah. You got me thinking about something to do my laundry now instead of just the fiber ends, but >> Yeah. Okay, I think that's a good way to wrap and yeah, I just want to thank everybody for watching and appreciate you coming in and taking the time. Thank you.