The Joy of Solving Complex Challenges ft. Krishna Govindarao of LightMetrics

[00:00:00] Now in the middle of one of the most significant eras in automotive, Carl Anthony amplifies the minds and voices behind this historic transformation as the host of AutoVision News Radio. Mic check 1 2

[00:00:11] All while coming to terms with middle age, father loss, and what it means to be successful in Detroit. As part of AutoSense and In Cabin USA 2024, we hosted a number of industry thought leaders as part of our AutoVision News live broadcast from the exhibition floor.

[00:00:29] One such expert was Krishna Govindarath, co-founder and head of product at Lightmetrics. Each day Krishna has the unique experience of working on complex and challenging problems in computer vision and machine learning, which he combines with over two decades of experience as an entrepreneur.

[00:00:47] Krishna has over 25 granted patents and is named as an inventor in over 50 patent applications. Lightmetrics is a provider of video telematics solutions for commercial vehicles, the most notable being its RideView platform for fleet applications.

[00:01:05] RideView is hardware agnostic and provides driver coaching workflows to increase safety and recorded video for accident exoneration among other features. Lightmetrics has made what it calls a strategic expansion by developing driver monitoring systems for passenger cars, leveraging key aspects of RideView in the process.

[00:01:27] One such aspect is how Lightmetrics can access billions of miles of driving data with different camera positions, diverse driver and occupant ethnicities, and varying road and weather conditions all thanks to RideView. I'm pleased to be able to replay for you now.

[00:01:45] My interview with Krishna, our guest during AutoSense and In-Cabin USA 2024. Moving at the speed of mobility, this is AutoVision News Radio with Karl Anthony in Detroit, Michigan. Lightmetrics is a nine year old company. We work in the video safety space.

[00:02:02] We work with commercial vehicle fleets to keep their drivers safe on the road. We detect risk based on how drivers drive, whether they're distracted, when they're fatigued, we alert them. And we also help fleets build report cards for drivers.

[00:02:18] This is not meant to be a punitive system, but it's meant to be more of a positive behavior reinforcement system. So we also provide tools for fleets to coach drivers, help them get better and stay safe. Absolutely. Krishna, what brings you here this year to AutoSense and In-Cabin?

[00:02:36] What brings you to Detroit this year? Sure. So when we talk about understanding risk of driving for commercial fleets, we do a lot of driver monitoring ourselves. And In-Cabin is the place to go to for driver monitoring. Sure.

[00:02:49] So we really wanted to see what else is going on in driver monitoring and to stay up to date. That's why we are here. Yeah. As you know, Krishna, DMS is a very competitive space. Why have you chosen, why is Lightmetrics chosen to jump into this space?

[00:03:09] Right. That's a great question, Karl. And we get asked that all the time. There are a few things. So we have been doing DMS in the aftermarket for some time now. So we are not new to DMS. We are new to this space with passcars. I see.

[00:03:23] What gives us some advantage, some advantages we have are we have a DMS tag that's thoroughly tested every day. And commercial vehicles drive a lot more than passenger cars do. Yeah. So our algorithms are really well tested for all sorts of coronal cases.

[00:03:38] We have very low false positives. So we are not starting from scratch here. The second thing is all DMS relies a lot on data. You go see any number of people here who are helping people with data to improve their algorithms. Of course.

[00:03:52] And we have access to tons of data because of our aftermarket deployments. So these are some of the things that make us very confident. Our experience, our AI stack and the data. That makes perfect sense. Krishna, you mentioned your AI stack.

[00:04:06] Tell us a little bit more about your AI stack. Thank you. So that's one of our things. So for example, my background today, I'm a head of product, but my background is in AI. And we have a few things that we do really well.

[00:04:20] The demo that we are showing here in the show is does your neural network know what it does not know? So for example, if you have trained a neural network to detect a cat or a dog and you show it a horse, a neural network guesses. Right?

[00:04:38] So it does not know what it does not know. What this means for car users are annoying beeps. Whenever it sees something that it does not know, challenging lighting situation, something else like a mask, my hand on my face or something like that, it goes crazy.

[00:04:55] That is one thing. The second thing, so able to being able to tell uncertainty. When I give a decision, I'll tell you my uncertainty, my confidence. That's a big deal in what we do. The second thing is our computational complexity.

[00:05:11] As the industry moves towards EVs and DMS is being integrated in mirrors and infotainment, computational complexity. How much power you consume is becoming a big factor and we are masters of low complexity algorithms and architectures. And lastly, we are really good with fine grain classification.

[00:05:31] So when people wear seat belts on similar clothing, similar colored clothing, those are the kind of places where you know, the AI actually can do almost as well or slightly better than humans.

[00:05:40] So we are able to do things like I'm holding a cell phone versus I'm just holding my ear. Right. Those are the kind of gestures where the annoying beeps can really annoy you. I see. These fine grain classifications are what we do.

[00:05:51] These are some of the things we do really well. That makes sense. Krishna, in your bio and I'll read this again. Each day, Krishna has the unique experience of working on complex and challenging problems in computer vision and machine learning.

[00:06:02] Krishna, what is it that you enjoy most about working on complex and challenging problems? What do you enjoy most about doing that? Okay, this may sound a little weird. One, I like solving complex problems for the sake of it.

[00:06:16] It's just, you know, it's like how you pick up a puzzle and solve it. You just get a lot of joy from it. But then what really drives us forward is the impact it has. Yeah.

[00:06:25] You know, like I said, we all know, right, all of us who bought cars with Lane Keeping Assist 10 years back, we just turned it off because it would keep dipping us at all the wrong times.

[00:06:36] So when you take hard problems and solve them and make it actually useful, that gives you a lot of confidence. In my previous kind of life before Light Matrix, I worked a lot on improving smartphone pictures. That was great.

[00:06:50] But the work I do now is a lot more satisfying. Absolutely. Krishna, from all of us here at AutoVision News and Sense Media, thank you for your time, sharing your expertise and in your story as well. Have a great rest of the show.

[00:07:03] And when the event ends, safe travels back home. I truly enjoyed this. Thank you. Sure. Thanks, Carl. Thanks to you and your team. Really appreciated you having me here. For more information about Light Matrix, see the links in the show notes.

[00:07:16] AutoVision News Radio is available on Spotify, Apple Podcasts, Podbean and more. From the Huntington Place in downtown Detroit as part of AutoSense and In Cabin USA 2024 alongside Krishna Govindarao. I'm Carl Anthony. AutoVision News Radio.