Get In: The Connected Vehicle Podcast From BlackBerry (Episode 8)
How many people – or companies – does it take to design a modern automobile? It may sound like the setup for a joke, but in reality, it’s no laughing matter. Today’s cars are marvels of complexity, the synthesis of myriad discrete systems that must come together harmoniously – and perform to perfection – while hurtling down the highway at high speed. That’s a lot easier said than done.
Welcome to the eighth episode of “Get In: The Connected Vehicle Podcast from BlackBerry.” This series explores the possibilities created by – and technologies behind – the revolution in global transportation we are witnessing today. In this episode, we meet Bhaskar Dani, Head of Cloud Alliances at KPIT, a leading company driving the shift toward the software-defined vehicle.
KPIT has carved a distinctive niche for itself as a master of automotive integration. In our podcast, Dani explains how KPIT specializes in choreographing the complex process of uniting the efforts of a global ecosystem of suppliers and engineers, and how the new generation of connected vehicles is challenging them to sharpen their skills.
Like most people in the automotive industry, Dani’s passion for cars started at a young age. But growing up in India, Dani came of age in a markedly different environment for motor vehicles than in the United States, where he lives now. “India is primarily a two-wheeler place, and I got on a bike before I got behind a wheel,” he says. Back then, only two car models were widely available in India: A Fiat, and the Ambassador, both of which are now discontinued. “My family had the Fiat, and that's where I learned driving on Indian roads. It was a very interesting experience. The gear shift was on the side of the steering column, there was no power steering, no power windows, no car radios, no AC. But driving was fun.”
Since moving to the U.S., Dani has seen his own car needs change as his life requirements evolved. As an engineer, he gravitated to designing many of the systems and “creature comforts” that his family’s Fiat sorely lacked. As his career progressed, so did the role of the infotainment systems he designed. In fact, it’s not a complete exaggeration to suggest that that much of the highly advanced technology in today’s software-defined, highly connected vehicles had its beginnings in those early infotainment systems.
“With my role in infotainment and car radios, I got to see the whole move from single-line LCD displays to display audio information, to today’s connected services and software defined vehicles,” he says. This has been an inexorable drive toward an increasingly user-focused experience for the driver. “Car makers have tried to personalize cars for a while now. From the features and functions that touch the customer directly – interaction and nowadays voice – there is even more demand for personal features for a car.”
This trend has become perhaps a fundamental force for change in the automotive industry. “Consumers want their cars to be the ‘latest and greatest’ at the point of purchase,” says Dani. But what's now changing is they also want those cars to remain up-to-date over their lifecycle.
“As their personal digital lifestyle changes and evolves with technology trends, they want the car experience to be continuous with that. The traditional software development systems and methodologies that were being used by car makers cannot address that problem fast enough.”
The solution has required entirely rethinking the focus of vehicle development. “It’s led to this whole initiative in the industry for what we are now calling the ‘software defined vehicle,’ which is fundamentally rearchitecting the car as a platform, and not as a fixed feature set that remains monolithic over the life of the car,” says Dani. “There's a fundamental change happening in the core software architectural foundation of the car. This is the ability to add adaptability and portability to the data and the applications, leading to a menu card of features that customers can pick, add, and subtract over the ownership of the vehicle – not just at the point of purchase.”
KPIT as a company is working closely with numerous automotive manufacturers to support and enable this fundamental design shift. Unbeknownst to most consumers, many “car makers were primarily system integrators, and the suppliers brought in multiple dedicated, one-function-only systems,” says Dani. KPIT made its name helping car makers integrate these disparate systems, which typically required copious amounts of custom software that KPIT provided.
“There is a significant shift in how software integration with third-party hardware is happening. And that's where standardization is coming in,” says Dani. “As car makers are building software more and more in-house, there's an element of integration now required at a scale which has not been seen before.” The trend of favoring in-house software development could have reduced the role of KPIT, but the company has instead seen growth in this part of its business. “Our traditional role of providing software to a set of requirements, testing it out and delivering at a certain quality – that is changing to a very loosely defined set of variables that are constantly changing.”
At the same time, KPIT has seen development cycle times come down to as short as 18 months. “KPIT is emerging as a player with the software integration capability that is bringing all of these variables into a common baseline ‘software factory,’ and delivering it as the production happens.” This means bringing together all the disparate elements of a vehicle’s software stack, from noise reduction and Bluetooth® infotainment-related features to over-the-air automotive updates.
This is also where the BlackBerry IVY™ automotive software platform is proving crucial. “One of the requirements of the software-defined vehicle movement is the ability to run applications and services, both at the edge – which is on the vehicle – but also on the cloud,” says Dani. “BlackBerry's hitting a lot of right notes with the IVY initiative.”
Overcoming Obstacles
“The major obstacle today is that there are too many variants in car sensors,” he says, from tire pressure monitoring systems, to cameras, battery, and any of the dozens or even hundreds of discrete functional domains that are instrumented in today’s modern automobiles. The systems, sensors and data they produce can vary greatly, “by OEM, by supplier, by program, and by geography. There's more time spent in standardizing those data sources and their formats, and then exposing them to third-party developers, than actually building the applications,” Dani says.
“IVY is the only initiative where there's an element of standardization across all of the sensors, across all of the types of car maker formats,” says Dani. “It abstracts the developers and also the car makers from having to worry about a car-by-car, program-by-program, variant-by-variant integration effort.”
This standardization can greatly accelerate time-to-market for new products and services, and removes a lot of challenges, too. However, as the amount of data produced by the vehicles explodes, securing that data becomes increasingly important. Managing that data efficiently and reliably has also become a critical safety consideration, particularly as features are integrated and consolidated among fewer electronic control units (ECUs) within the vehicle. With discrete systems, if one crashes, the others will carry on. But with greater integration, it’s important to ensure that a failure in one area doesn’t affect others.
Exploring “the Next Big Thing”
KPIT has been exploring what new services can be made possible by the integration of vehicle systems. One of these, called Connect, allows multiple device users in the car to interact with the vehicle in interesting and useful ways, such as controlling the music playback, or sending an address to the car’s navigation system. “In order to do that, we built a many-to-many platform that can bring in any device. It could be a phone, a tablet, a garage door opener, and all sorts of heterogeneous devices, but integrating them with the car in a very homogeneous, single-API fashion.” KPIT has also created a solution where actions of the car affect parameters of a game an occupant is playing, perhaps via a virtual or augmented reality headset. “One of the features we built on similar lines was taking a feed from the camera on the car and feeding it wirelessly through to the phone. There's another one that we built where the instrument clusters, the speed and the gauges are replicated on a smartphone display.”
These capabilities fall under the heading of vehicle-to-everything connectivity, also known as V2X, which also brings interoperability with cloud-provisioned services into the frame. This can have further “knock-on” benefits for system development. “One of the major trends in this is virtual validation,” says Dani. “The entire ECU at a functional level is getting virtualized and being hosted in the cloud. With that, you can remove dependence on hardware availability, and make that whole development process global. Initially people thought cloud was primarily about delivering services, but it's making a very big change on the product development side as well, especially in software-defined vehicles and other connected autonomous pieces.”
Chips Aren’t the Only Shortage Affecting the Auto Industry
The key to taking advantage of this opportunity is having enough of the right software developers on board. “Those car makers that can invest and build teams with thousands of engineers for doing the software are already doing so, and those that aren't are already on their way to partner with others who can bring that,” says Dani.
While there are plenty of developers able to generate apps and the content cloud, the emerging area of autonomous systems is a different matter,” Dani says. “For the entire development cycle from data ingest to the modelling, training and then simulation validation, this is a particular feature set.” These are not common skills for software developers, and it’s creating a talent shortage that is becoming increasingly evident across the automotive industry.
“The second challenge we are seeing is car makers saying that they are still not where they would like to be in an integrated vehicle-level validation cycle,” says Dani. This is where developers like KPIT and platforms like BlackBerry IVY can smooth the process of developing for the software-defined vehicle. “We are going to see a real step up in refinement of how things come together in the car experience in the coming years.”
You can subscribe to "Get In: The Connected Vehicle Podcast from BlackBerry" at the following media services:
Video Transcript Steve Kovsky: Bhaskar Dani: Steve Kovsky: Bhaskar Dani: We basically got in the car for the experience of just driving. Over time, as my education career progressed, I came to the U.S. to study and settled here. I went through that curve of, value for money cars, and family cars. Then I had one exciting car, then a dog-friendly car. And this year my daughter will start driving. I'm probably going to be in the market for a highly safety fortified electric car just in about a year. It's been an interesting journey and it's purely a stroke of luck that my profession ended up in the same industry. Like I said earlier, with my role in infotainment and car radios, I got to see the whole movie – from single line LCD displays to display audio to projection to connected services and software defined vehicles. So it's been very interesting relationship with the car industry. Steve Kovsky: Bhaskar Dani: And by that, it means as their personal, to your point on personalization, as their personal digital lifestyle changes and evolves with technology trends, and initially we had content and apps and then favorite music. Now it's a lot more preferred devices and feature that are coming in. They want the car experience to be consistent with that and continuous with that. And that is leading to tremendous proliferation in the amount of software variations or variants that you would launch, and the traditional software development systems and methodologies that were being used by car makers, they cannot address that problem fast enough, they cannot catch up. And so that has led to this whole initiative in the industry for what we are now calling software defined vehicle, which is fundamentally rearchitect the car as a platform and not as a fixed feature set that remains monolithic for 10 years or over the life of the car. And to be able to build a platform like that, there's a fundamental change that is happening in the core architectural foundation of the car, software architecture foundation, ability to add adaptability and portability to the data and the applications, and all of it leading to almost like a menu card of features that customers can pick, add and subtract over the ownership of the vehicle, not just at the point of purchase. So, it's very interesting question you asked, and it's actually linked to one of the most fundamental changes in the industry. Steve Kovsky: Bhaskar Dani: As suppliers would create these, one function or multifunction systems, we would engage with them as well as some of the OEMs in building that software. And there would be either a software development activity going on, where we would take the features and build that software, test it out and deliver to the supplier. The supplier would then integrate that into the hardware integration, deliver to the car maker. Car maker has their own cycle of full integrated system test. And then that will go to production and launch. Over time, and by that, I mean, in the last, I want to say about 10 years, the balance is shifting. So more and more of the software portion of this activity is being brought in-house by the car makers, the OEMs as we call them. With that transition, what is now starting to happen is a pretty significant shift in how the software integration with third-party hardware is happening. And that's where the standardization is coming in. We've been automotive company for more than 20 years, and very well reputed for providing both solutions and services in multiple subsystems of the car. We had expertise and powertrain body chassis, of course, infotainment, diagnostics, automotive architecture, AUTOSAR as we call it. But it's now all coming together. And as car makers are building the software more and more in-house, there's an element of integration that is now required at the scale which has not been seen before. Our traditional role of providing software to a set of requirements, testing it out and delivering at a certain quality, that is changing to a very loosely defined set of variables that are constantly changing. And OEM is defining the requirements of what car they want to launch, in parallel with the development of those requirements, which are in parallel with the multiple software components that are coming from a variety of outside suppliers, who are also looking to launch their software component versions. And so those components are also in development and the OEMs need to cram this whole development cycle from traditional three years to 18 months to even six months at a feature level that's causing tremendous, I want to say that's really galvanizing how it's done. KPIT is really emerging as a player with the software integration capability, with the knowledge and domain expertise of each of those subsystems and being able to run software factory bringing all these variables into a common baseline and delivering it as the production happens. It’s interesting how our role has changed, and that's going to be our positioning as a software integrator, as the industry evolves into these new connected autonomous and electric vehicles. Steve Kovsky: Bhaskar Dani: But there was also an independent set software providers. Somebody was providing the echo cancellation noise reduction. Somebody was providing a Bluetooth® stack. Somebody else was providing an over the air update stack. Those things work really well in isolation, but their implementation on a particular platform for that OEM's requirement is a distance that has to be traveled for every single program every time. KPIT came in as a software integrator. Not only did we have our own core platform, but we actually ran the baseline software across multiple geographies worldwide, across multiple SOPs, or start of production dates, and actually integrated on a weekly basis then on a monthly basis and upgraded set of software versions that is getting validated in parallel as this software is getting developed. It's a very, very dynamic process. The science part of it is easy to understand and easy to replicate. It can get the right tools, get the right DevOps and other practices that large companies use. The art portion of it is where KPIT shines, that cannot be accelerated. That's just years and years of doing it and knowing what's the right validation methodology? At what granular they would want? So eventually this program actually went in production. At the time of launch it was the OEMs, I think it was the highest quality program at launch. To your original question, yes, there was a chance that companies like us role would shrink, but actually OEMs have seen that as the software variants and the components, and the requirement, and the speed of it is all changing and coming together, they need a software tier one, if you will, that's sitting next to them. That is bringing all the software practices, best practices of software development at large scale and at automotive rate. KPIT is actually seeing growth in that part of our business now. Steve Kovsky: Bhaskar Dani: BlackBerry is hitting a lot of right notes with this IVY™ initiative. The major block today comes down to the car sensors – tire pressure monitoring or what have you, by OEM, by supplier, by program, and by geography, there are too many variants. There's no standardization. And there's more time spent in standardizing those data sources and their formats, and then exposing them to third-party developers, than actually building the application. And even if that is done by a given car maker, they have tried app stores or equivalent approaches but haven't been very successful so far. What's happening with IVY is, it's the only initiative where there's an element of standardization across all of the sensors, across all the types of car makers’ formats. It (IVY) abstracts developers and car makers from having to worry about car by car, program by program, variant by variant integration efforts. Once that's there, the ultimate vision of making a car available to multiple third-party developers as a platform, to launch new ideas, new use cases, new services, new applications, is possible and accelerated. I think it's coming in at a very interesting time in the industry, and if anything, it'll accelerate the adoption of the car as a platform by the developer community. That has been a challenge in the past for the car industry, because of the sheer volumes of phones versus cars, but an initiative like this will accelerate that and take out a lot of challenges in the process. Steve Kovsky: Bhaskar Dani: If I'm trying to skip a song and I'm frustrated that it's not getting skipped, I start hitting the screen multiple times, and then it hangs the infotainment head unit. Traditionally, if this was an entirely separate system, it could reboot. With the consolidation on central compute, things are different now. One subsystem crashing or rebooting can affect the safety critical systems such as the instrument cluster, which is supposed to show you all sorts of real-time data. This one element is getting very critical. And of course, you and I read about cyberattacks. That's the other element. The foundational software defined vehicle views a few things like, upgradability, adaptability, and so on. It's all under the umbrella of safety and security. Steve Kovsky: Bhaskar Dani: The cognitive overload is a real deal. A lot of that being taken over by voice systems and AI is going to be a major change. Today it's better than it was yesterday but it's still not at the level where it can go. That's a major change we're going to see. The other thing which is a little bit far away is 5G and eventually 6G type. Low latency communication between vehicles and infrastructure. That's something that's going to come, it's a little bit far out today. On the safety side, of course, level three AIDA functions are starting to come in most mainstream vehicles. That will only increase. From the autonomous standpoint, until people really experience how good that is, (only a small portion of the market has actually seen how it feels and how comfortable it is), once it becomes mainstream, we'll be able to see how it evolves. Steve Kovsky: Bhaskar Dani: This was one example. To do that, we built platforms that basically bring many devices that are not consistent; a phone, a tablet, a garage door opener, and what have you, all sorts of heterogeneous devices, but integrate them with the car in a very homogeneous, single API type fashion. It was very, very versatile solution. To push the limit on that, we went ahead and built an in-car gaming experience. We brought in a game studio essentially, once you have a secure point to point way to exchange data between the car and a third-party device, you could stretch the limit. So that's what we wanted to do. Essentially, this is a new group of use cases that are starting to come on the horizon, which is customers or car occupants bringing their own mobile devices, or we see almost even virtual reality headsets that are integrating with the car through a platform like ours. At that point, based on the real-time braking, accelerating, cornering that, perhaps your dad is doing, you could be a kid in the back playing a game and the parameters in the game are affected in real-time. So that's something that we proved out. It didn't reach production yet. We are continuing to do many things around that. That's an example of a use case that we've actually launched. The remote control piece is already in production for a few years now. And very, very appreciated by both users of... the drivers, as well as the passengers. Steve Kovsky: Bhaskar Dani: Steve Kovsky: Bhaskar Dani: Then you need over air updates and downlink making that learning part of the entire fleet, so that next time a vehicle encounters that scenario it knows how to handle it. And an extension of that is vehicle to vehicle communication, and then vehicle to infrastructure communication. Those things are going to expand the realm of possibilities significantly more – and will need even more software integration. Steve Kovsky: Bhaskar Dani: One of the major trends in this is virtual validation. Car makers with the amount of software that's coming in, and the amount of the sheer scale of it has reached a point where if they want to shrink the cycle time of development from requirements to launch, they cannot wait on first getting the chip makers to design the right chips, then the tier one supplier is building hardware prototype, then testing that, and then waiting six months later finding out that there are these issues and the old sequential product development, they cannot afford to go there. So now what's happening is the entire ECU as we call it, or at least at a functional level is getting virtualized, and that's being hosted in the cloud. With that, you can remove dependence on hardware availability and testatrix availability and make that whole development process global. You can go where the talent is, put it all on cloud and parallel development. Thousands of engineers plugging into one single platform, building an entire vehicle unified software that has become a reality now. It's a new trend, it is happening. OEMs are making an investment, but we are seeing something that is unprecedented. Cloud is enabling that. The other piece cloud is enabling is the services that are getting delivered. Having a replica of the car in real-time, that's been tried out in manufacturing under the digital twin terminology, but for cars with the dynamic system, with so many more variables, it's going to be very interesting. And without cloud-based data collection and intelligence, I guess, generation with AI and distributing that, it's not possible to give that experience. The cloud is making a very, very fundamental change. Initially people thought it was primarily about delivering services, but it's making a very big change on the product development side as well, especially in software defined vehicle and other connected autonomous pieces. Steve Kovsky: Bhaskar Dani: There's apps and content cloud development, which is there, but the core engineering of it, if you take autonomous development, if you see the entire life cycle or the development cycle from data ingest to the modeling, training and then simulation validation, this is a particular feature set. It's not general software, general IoT, general IT, in fact. And so one major trend is bringing the right amount of talent fast enough. The second challenge we are seeing is car makers saying that they are not still at where they would like to be in an integrated vehicle level validation cycle. What I mean by that is they have great integration and focused software test validation frameworks, as well as development methodologies for each subsystem in the traditional development cycle. Everybody in a global OEM will say, this part works infotainment, this works great in our lab, with an integrated vehicle, we don't know. The brake systems, the chassis systems, the steering systems and so on. And it is on the OEM, especially the teams that launch that software, it is down to them to really bring it all together and sign up for production worthiness of the software we are releasing. The number of releases are increasing, the complexity of software that's coming together is increasing. The time to validate is even lesser. And so that's a challenge where some new frameworks and new methodologies are coming. We have one ourselves, we have looked at these problems, hearing from the OEM, we've looked at these problems multiple times. And as I said earlier, having gone through building those solutions, I do think that if we are going to see some really step up in refinement of how things come together in the car experience in the coming years. Steve Kovsky: Bhaskar Dani: Steve Kovsky: |