Data powers GenAI to develop more intelligent EV functions

GenAI is emerging everywhere, but Thomas Mueller, VP, CTO and Global Automotive Lead at Wipro, believes it will excel in EVs with a suitable approach

Perhaps the hottest topic of 2023 was generative artificial intelligence (GenAI) and it’s almost impossible to visualise an aspect of life or business unsupported by such a highly intelligent technology. With electric vehicles (EVs) becoming increasingly intelligent, there’s no doubt that GenAI will play a key role in enabling autonomous driving, intuitive cabin experiences and reducing the cost of driving. 

Uncovering the use case for GenAI in the most intelligent vehicles the industry has ever seen, Thomas Mueller (TM), VP, CTO and Global Automotive Lead at Wipro, wasn’t short of a few insights into this topic. Having sat down with him for an extensive conversation on the subject, two things became clear: Firstly, GenAI is already here and in use in automotive. Secondly, Mueller is incredibly passionate about the future of this technology. 

TS: Who you give us an outline of GenAI’s integration into EVs? 

TM: First of all, the powertrain itself does not limit the use of AI, slash generative AI in automotive. Technically speaking, we have been using AI and machine learning quite successfully for over a decade to embed technology in automotive. Particularly everything that you see today, including driver assistance technology, leverages machine learning and AI models in some shape or form, and the evolution now to what has been the generative part of it. That's an interesting transition.

With electric and hybrid, you obviously have the inevitable range anxiety that people carry forward. Range anxiety can only be addressed by putting in bigger batteries, which counter feeds the purpose.

We have the ability to bring what we call intent and context into the product with AI that sits in the vehicle and it sits outside the vehicle. This is not specific to a car, it can be a truck, van, or even a two-wheeler. If the product has the ability to aggregate information pertaining to its users, whether it's the driver or its passengers, we can improve the overall user experience and the product characteristics.

TS: How does that work exactly? 

TM: We have, what we call, stretch knowledge graphs. Knowledge graphs are semantic graphs that have been used in AI for quite some time that we can stretch on our cloud platforms that sit inside and outside the vehicle. 

The first evolution is the in-vehicle architecture. Historically, this has incorporated monolithic, spaghetti-like code; embedded C code and, unlike everything else in the industry bar a few, has been evolving into service orientation—delivering software in microservices and containers for over a decade now. The modularisation of software has not happened, particularly in the automotive space. We are, first of all, disaggregating the vehicle software, making software and hardware independent from each other.

We can deliver them in containers and, once we have these containers, we can place them on a platform that connects to a public or a private cloud. Tesla is using a private cloud, notably. It's interesting to see that everybody else uses the public cloud. The market leader is not. I believe there's some really good reasons for that, especially when it comes to the capabilities of AI.

Once the platform is ready, that software becomes a mechanism to constantly change and grow—we can deliver data in and out of the car in a continuous manner. We are basically creating what we call a loop or a data engine that extracts data and brings model updates to the vehicle in a continuous manner.

TS: How is data used specifically for AI applications in EVs? 

TM: When it comes to data, the question is: “Is the product able to separate between the bulk of data that it generates from sensors?” They just take the in-cabin monitoring solution. So if you think of cameras that are pointed at passengers and drivers, they can do a number of things. You can always detect anomalies, someone smiling, someone making facial gestures, someone falling tired. Everything that you can point cameras at inside and outside the car generates a huge amount of data. 

Most of the data is worthless, but car makers—excluding Tesla—lack the ability in real time to separate the valuable from non-valuable data. That's the first capability, uplifting that comes if you're using powerful chips, powerful computers. Today those computers are mostly exclusive to a premium product with all the options like, for example, an S-Class Mercedes, which has the driver assistance option if the user has the larger computer. If you don’t opt for this package, the computer is not in the car. 

TS: Explain how GenAI revolutionises the in-car experience.

TM: You can optimise the cabin comfort to align with efficiency in a way that between surface heating—like heated seeds, steering wheels, armrests and other surfaces that can be heated and cooled—you can use that compared to just heating the air in a vehicle, which is very wasteful.

Every time you open the doors, you lose energy from heat and that's a scarce resource. So we can optimise this information continuously along with the experience of the product—that's just a few things assimilating how you drive and learning and reinforcing this—it goes another way. The generative part of that comes in when you can use different functions.

When you want a new feature in a car, it's typically down to very few people in the coding department who can code exclusively. And if you are a business owner who says, “I want to change my app in the car from a blue colour to a green colour”, that simple change can take you months if not quarters, and costs you a lot, and you have to go to the back of the line of your coding department or coding partners to get that change done. 

What we typically do is provide a spare capacity on the supercomputer, which is used for resilience and failover. Just in case the computer fails the assisted driving functions, it needs a backup computer at least when we get to level 3 autonomy.


Make sure you check out the latest edition of EV Magazine and also sign up to our global conference series - Sustainability LIVE 2024

EV Magazine
 is a BizClik brand 


Featured Articles

Voltpost's Lamppost EV Charging can Electrify Urban Mobility

Determined to electrify urban mobility, Jeff Prosserman, CEO of lamppost EV charging company Voltpost, shares how his team is democratising charging access

Battery Innovation & Female Leadership in China's EV Sector

The EV landscape in China see sales surge amid battery innovation & women in leadership roles, who restructure supply chains & navigate geopolitics

Alstom's Electric Trains lead Sustainable Growth in Perth

Alstom's AUD$1bn contract for unit C-Series electric trains will ensure commuters in Perth can enjoy a clean & punctual commute, as e-trains gather speed

NIO Launches Smart Driving Technology Center in Germany


Tesla & VinFast rev up EV manufacturing in India


Aggreko to Support Formula E’s Sustainability Targets