Latest / eMotors: Electric Revolution / AI and Robotics Driving Perfect EV Reliability
Transcript
- 0:09Electric vehicles are, I mean, they're undeniably the future of
- 0:12how we get around. Oh, absolutely.
- 0:14But. They bring this new level of
- 0:15complexity, right? Especially when you think about
- 0:17those huge battery packs. It's something traditional car
- 0:20manufacturing just never had to deal with.
- 0:22That's the key issue. So if we want EVs to be truly
- 0:26reliable, to be safe, to last, the quality can't just be good,
- 0:31it has to be basically perfect. Precisely.
- 0:35The goal has shifted. We're not talking about
- 0:37acceptable error rates anymore. We're talking about 0 defects 0.
- 0:42And the stakes are just monumental.
- 0:44The market is absolutely surging.
- 0:45We saw over 10 million units sold back in 2022.
- 0:50That was a 55% jump from the year before.
- 0:53So to keep that growth going, manufacturers need a degree of
- 0:56precision that conventional assembly lines just can't
- 0:59deliver. And that all comes down to the
- 1:01battery, doesn't it? It absolutely does when you're
- 1:03dealing with the chemistry and the structure of a modern
- 1:05lithium ion battery. Everything performance, safety.
- 1:08It all hinges on tolerances measured in micrometers.
- 1:11Micrometers just for context. How small are we talking?
- 1:14A micrometer is 1 millionth of a meter, yeah, so think smaller
- 1:19than a single particle of dust. If you miss your alignment by
- 1:23that much, you risk serious performance loss.
- 1:27Or, you know, worst case, a major safety event.
- 1:30OK. That really puts the challenge
- 1:32in perspective. So let's untack this.
- 1:34Let's do it. Our mission today is to do a
- 1:36deep dive into how AI and robotics are tackling this very
- 1:40problem, turning the electric car into, well, a product of
- 1:44predictive intelligence. And we've seen some data
- 1:47suggesting these technologies are boosting reliability,
- 1:50reducing failures by somewhere in 20 to 30%.
- 1:54And that's not all. They're also cutting downtime,
- 1:56both in the factory and for the vehicle itself, by about 50% on
- 2:00average. 50% that's seismic for a manufacturer.
- 2:02It's complete transformation means faster production, lower
- 2:05costs and crucially, a much higher quality car coming
- 2:09offline. It's all driven by intelligence
- 2:11and automation. So let's start with how this
- 2:14impacts you, the driver, because we tend to of robots just on the
- 2:17assembly line, right? At building the car.
- 2:19But this tech plays a dual role. It's perfecting the car's
- 2:22creation in the factory, and then it's actively working to
- 2:25improve and maintain it while you own it.
- 2:27That dual role is so important. The car itself is becoming this
- 2:32intelligent asset after it's sold.
- 2:35The onboard AI is running these really sophisticated predictive
- 2:40diagnostics. So it's not just waiting for a
- 2:42warning light to pop on. Not at all.
- 2:44It's continuously analyzing real time sensor data from everything
- 2:49the battery, the powertrain, the brakes, and it's looking for the
- 2:52tiniest anomalies. So the car is basically giving
- 2:54itself a constant health checkup.
- 2:56Exactly. Looking for those subtle stress
- 2:58signals before they ever become a problem you'd notice.
- 3:00And it's not just about warnings, it's about
- 3:02personalized interventions. What do you mean by that?
- 3:05Well, think about charging. The AI can customize the
- 3:08charging cycles, control the power draw, and optimize your
- 3:11range based on your specific driving style.
- 3:14And where I live, the weather. That kind of thing.
- 3:16All of it. And this proactive management
- 3:19prevents the battery and other components from aging
- 3:21prematurely. Can you give us a real world
- 3:24example of that? Sure.
- 3:25Look at Porsche. They're using machine learning
- 3:28for what they call predictive aging models, and they credit
- 3:31that with boosting the overall battery lifespan by a reported
- 3:3520%. 20% is huge. It is.
- 3:39Or look at Renault. They're using predictive
- 3:41diagnostics along with battery certification programs, which
- 3:45gives owners real Peace of Mind about the health of the most
- 3:47expensive part of their car. That's fascinating.
- 3:50It really changes the whole ownership model.
- 3:53But you know, even with all that AI, things still wear out.
- 3:56The service shop isn't going away.
- 3:57No, but even the maintenance Bay is getting a a 4.0 upgrade.
- 4:01Oh, Fasim 4.0 right, Workshops 4.0.
- 4:04Exactly. These places are starting to use
- 4:06advanced robotics, specifically something called Cobot's
- 4:09Collaborative Robots. So what's a cobot doing in a
- 4:12pear shop? Is it just tightening bolts for
- 4:13the mechanic? It's way more than that.
- 4:15I mean, they're designed to work safely right alongside the human
- 4:18mechanics, helping with heavy or delicate tasks, but their real
- 4:21value is an autonomous diagnosis.
- 4:24A cobot can scan a complex battery pack, use precision
- 4:28imaging to find the exact failure point, and then help the
- 4:31human mechanic make a really fast, error free repair.
- 4:34So it just speeds everything up. Massively.
- 4:37And since the car is connected, it can essentially self plan its
- 4:40own service appointments based on those predictive alerts.
- 4:45The sources we looked at suggest this can extend the useful life
- 4:47of an EV by anywhere from 15 to 25%.
- 4:51That completely changes the economics of owning one.
- 4:55Now let's pivot to the production side, because this is
- 4:57where things get really, really interesting.
- 4:59OK, there's a huge philosophical shift happening inside the
- 5:02factory. We are moving away from what we
- 5:05call reactive quality control. Which is just finding mistakes
- 5:08after they've already. Happened, right?
- 5:09And then you have to scrap the part.
- 5:11Instead, we're moving to predictive manufacturing
- 5:13intelligence. Prevention over cure.
- 5:15Precisely the ultimate goal is 0 defect manufacturing.
- 5:19It's this ambitious, almost utopian idea where every single
- 5:23step is optimized to eliminate defects entirely.
- 5:26But isn't 100% perfection just economically impossible?
- 5:30What's the trade off? Historically, yes, it was way
- 5:33too expensive, but AI and robotics are changing that
- 5:37equation for assembly tasks. Robots aren't just faster, they
- 5:41are exponentially more consistent, especially with
- 5:44those complex repetitive jobs and battery assembly.
- 5:48That precision is why some manufacturers are reporting a
- 5:51reduction in human related errors by a staggering 90%. 90%
- 5:56That's the kind of number that lets you scale globally without
- 5:59quality falling off a Cliff. It is.
- 6:01And our sources point out this is fueling an intense
- 6:04international competition. It's almost like an automation
- 6:07arms race. Arms race, Yeah, the level of
- 6:09automation is just staggering. Some factories, especially in
- 6:11China and with companies like Hyundai, are pushing toward
- 6:14being almost fully AI robotized. And that push for hyper
- 6:18efficiency can be, as one source put it, terrifying for legacy
- 6:21automakers. I can imagine and Tesla's in on
- 6:23this too right? Using their own AI.
- 6:26Absolutely. They're integrating everything
- 6:28from the Autopilot software to their Optimus robot to unify how
- 6:32they build and diagnose their vehicles.
- 6:34Let's go back to the battery for a second.
- 6:36You mentioned those micrometer mandates.
- 6:39This is where the human eye just can't keep up, right?
- 6:41Exactly like we said, lithium ion cells need electrode
- 6:45alignment within micrometers. If that alignment is off by just
- 6:48a tiny fraction, you can compromise the cells integrity.
- 6:52And that leads to short circuits.
- 6:53You can, yeah. And with the power density in
- 6:56these batteries, a microscopic defect can escalate into what's
- 7:00called thermal runaway. That's the chain reaction, the
- 7:02fire. That's the catastrophic failure.
- 7:04And that's why you need technology that can literally
- 7:07see and think better than a human can.
- 7:09Which brings us to the core technology making this all
- 7:11possible. AI enhanced machine vision.
- 7:15OK, we're essentially teaching the factory floor to see and
- 7:18think with a clarity and speed that is just far beyond human
- 7:22capability. It's the nervous system of this
- 7:24whole revolution. So what makes these vision
- 7:26systems so powerful? It's more than just a camera
- 7:29snapping a picture. Oh, much more.
- 7:31They're incredibly sophisticated multimodal systems that use high
- 7:35res imaging thermal monitoring. They combine data from visible
- 7:39light, infrared, sometimes even X-rays.
- 7:41So they can see things that are physically invisible to us.
- 7:44Correct, but the real magic is the AI.
- 7:47The deep learning algorithms are what make them intelligent.
- 7:50They can tell the difference between a true critical defect
- 7:54and just an acceptable little variation from the manufacturing
- 7:57process. So let's talk specifics.
- 7:59I assume checking the battery cells as they're made is job
- 8:02number 1. It's the single most critical
- 8:04application. Take electrode alignment.
- 8:07This happens at high speed. The system uses high res cameras
- 8:10and laser displacement sensors to verify positioning down to
- 8:13the micrometer. And that check is instantaneous.
- 8:16Instantaneous, and it's what prevents those internal short
- 8:19circuits that cause so many failures.
- 8:20OK, so the cells are aligned, now you have to package them.
- 8:23What about welding? That's where Weld quality
- 8:25inspection is critical. You use laser welding to
- 8:28assemble the battery packs, and the vision system uses real time
- 8:31thermal imaging during the Weld itself.
- 8:33Not after. Not after it monitors the Weld
- 8:36pool dynamic as they happen to spot things like incomplete
- 8:39fusion or heat affected zones that a later check would totally
- 8:44miss. Monitoring the heat in real
- 8:46time? That seems like the best
- 8:48possible defense against future thermal problems.
- 8:51What about the fluids? The electrolyte filling.
- 8:53Absolutely. Vision systems monitor that in
- 8:56real time, too. They're looking for tiny air
- 8:58bubbles or incomplete wetting of the components, and then they
- 9:01use thermal imaging during the battery aging process to find
- 9:04any hotspots that could indicate an internal weakness.
- 9:06It's just layers and layers of monitoring.
- 9:09It's the only way to get to near perfect quality.
- 9:11And the numbers coming out of the companies doing this are
- 9:13just incredible. We're seeing real, measurable
- 9:16winds. We are Tesla's Gigafactory for
- 9:18example, started using these vision driven digital twins.
- 9:22A virtual model of the factory. Exactly.
- 9:25And they reported a 34% reduction in assembly defects
- 9:28and a 28% jump in what's called overall equipment effectiveness,
- 9:32or O EE. O EE is a term we hear a lot for
- 9:36anyone listening. What does that actually mean?
- 9:38OEE is just a metric for how well a factory is being used
- 9:42compared to its full potential. It combines quality, performance
- 9:46and uptime. So a 28% jump means they are
- 9:49making way more high quality cars faster with less wasted
- 9:53time. And the really elegant part is
- 9:55the closed loop system where the factory corrects itself
- 9:58instantly. That's the goal.
- 10:00The vision systems aren't just finding errors, they're the
- 10:03feedback mechanism. These systems can monitor, say,
- 10:06up to 127 critical variables, and if one deviates, they adjust
- 10:11production parameters. In how fast?
- 10:130.3 seconds. That's faster than you can
- 10:15blink. It is, and that responsiveness
- 10:18is what creates consistency. Look at BY DS Blade battery
- 10:21line. They achieved 99.8% dimensional
- 10:25accuracy and cut material waste by 31%.
- 10:29That waste reduction alone is a massive cost saver, and this
- 10:32ability to see and predict it has huge strategic value that
- 10:36goes beyond just the car it affects the factory's own.
- 10:39Health. You mean predictive maintenance
- 10:41for the equipment? Exactly.
- 10:42So they're predicting when the robots might fail, not just the
- 10:44parts they're making. That makes sense.
- 10:46These vision systems are monitoring the machinery,
- 10:49processing something like 450,000 images a day.
- 10:53General Motors reported a 52% reduction in unplanned downtime
- 10:57at their EV plants after they put these systems in. 52% That's
- 11:02the difference between a small problem and shutting down an
- 11:05entire line for a day. Precisely.
- 11:07The AI is so good it can predict a bearing failure up to 72 hours
- 11:11in advance. So maintenance can schedule the
- 11:14repair during a planned stop and.
- 11:15What about traceability for recalls and things like that?
- 11:19Vision is critical for that too. Using high speed OCR and barcode
- 11:23reading, the system links every single part, every cell, every
- 11:27module to a digital record. Full audit trail.
- 11:30A bulletproof one. If a defect shows up years
- 11:32later, they know exactly which machine, which batch of
- 11:35material, and what parameters were used to make that specific
- 11:38part. It's essential for compliance.
- 11:40You know that focus on leveraging your own internal
- 11:42tech is becoming a strategic play in itself.
- 11:45We saw a really interesting move from Rivian on this.
- 11:47Right, the younger EV company. They had some real challenges
- 11:51early on, high defect rates, lots of scrap material on their
- 11:54battery lines, and they're trying to fix it all with better
- 11:57data analysis. And that internal effort to
- 11:59solve their own problems led to a fascinating strategic spin
- 12:03off. Mind Robotics.
- 12:04Exactly. Rivian launched it as a separate
- 12:07company focused just on AI and robotics for manufacturing and
- 12:11secured $115 million in funding. It's the Tesla playbook, isn't
- 12:15it? It is.
- 12:16They're taking all that operational data they gathered
- 12:18while fixing their own factory and using it to build smarter
- 12:22robotic systems that they can then sell to other industries.
- 12:25It's a brilliant way to create a new revenue stream outside of
- 12:28just selling cars. And it positions them as a true
- 12:31tech company, not just an assembler.
- 12:34Looking forward, this technology is only going to get sharper.
- 12:37We're already seeing things like hyperspectral imaging in labs.
- 12:40Which can see what exactly. I can detect subtle variations
- 12:44in the material composition itself, things that are
- 12:47invisible even to today's high res cameras.
- 12:49So the system will know if the chemistry is off, not just the
- 12:52shape. Precisely, And the robots
- 12:54themselves will get smarter. The future is autonomous vision
- 12:58guided robotics. They can adapt on the fly to
- 13:02tiny variations on the line without needing a human to
- 13:05reprogram them. And the goal is?
- 13:07The goal is audacious defect detection over 99.9% accuracy
- 13:11while cutting production costs by around 45%.
- 13:14It is a complete revolution. So what does this all mean for
- 13:17the person listening to us today, For the driver, the AI
- 13:22that's monitoring your car's battery, and the autonomous
- 13:24robots that built that battery? They're converging.
- 13:27They're creating one seamless self optimistic system.
- 13:31The next EV you buy isn't just electric.
- 13:33No, it's a product of predictive intelligence designed from the
- 13:36ground U for near perfect reliability.
- 13:38And this raises a really important question for the
- 13:40future. As manufacturers get to these
- 13:42near 0 defect rates with automation, what new skills will
- 13:46engineers need? Right?
- 13:47What do you do when you have to manage a system that essentially
- 13:49fixes and optimizes itself? The job becomes less about hands
- 13:53on work and much more about high level data interpretation and
- 13:56systems management. That is fascinating food for
- 13:58thought, watching the human role evolve alongside the tech.
- 14:01Thank you for joining us on this deep dive into the microscopic
- 14:05precision of the electric revolution.
- 14:06My pleasure. As always, we are so grateful
- 14:08for you, our loyal listeners. See at the 4T, if you enjoyed
- 14:12learning how AI sees the future of manufacturing, please make
- 14:15sure to follow the channel you know Fernom Pedersi, La Novita.
- 14:18We'll catch you on the next deep dive.