Why is AI so important for the automotive industry?
Jan Zawadzki is Head of Artificial Intelligence at CARIAD, leading the transformation of the car into a self-learning device. In this article, he explains the importance of AI, the challenges and opportunities it presents, and what we at CARIAD are doing to make the most of this technology.
Artificial intelligence is a technology that works by learning from data and finding patterns. With these patterns, it can develop predictions. And with these predictions, we can create new products, improve existing products, or lower costs by reducing complexity. Today, with huge amounts of data available to us, as well as more computing power thanks to better, state-of-the-art hardware, the time is right to harness artificial intelligence and put the breakthroughs that we’ve made in research into practice.
CARIAD is a software company and AI is a software topic. To us, AI represents the next level of software development. Instead of having programmers actively list rules and write code, we can train AI to learn code from data. Our mission is to transform automotive mobility, and that includes turning the car into the ultimate, self-learning mobility device to make the driving experience safer, more sustainable, more comfortable and more enjoyable. With the help of AI, the software and hardware platform that we’re working on will become the brain – the neural network – of this continuously learning fleet of vehicles.
How do we turn the car into a self-learning device?
As part of the Volkswagen Group’s NEW AUTO strategy, CARIAD will connect 40 million cars by 2030. We see a huge amount of potential here – that is, the potential to use one of the largest data-collecting fleets in the world to train AI for a variety of use cases. Use cases that we can’t even image today. With access to the data of those 40 million cars, we can create true customer value.
But how exactly do we go about creating self-learning cars? Firstly, it’s important to approach AI in a holistic manner. By doing so, we can truly make use of the technology across the entire Volkswagen Group, and the individual brands and functions. We need to work together in a community, with decentralized AI hubs across various domains, such as ADAS and intelligent cockpit. We need to share knowledge in a transparent way so that we as a company can better understand customers’ needs.
Secondly, CARIAD has implemented a comprehensive AI strategy. In this strategy, data is the underlying driver of growth. Every day, a single car generates 50 terabytes of data. But there’s no use throwing all this information into a data lake and telling data scientists to work on it. They’re going to drown. We need an intelligent way not only to collect the data, but also to work with it. The Big Loop is part of this process.
And thirdly, we need to pay very close attention to safety. If you’re using AI in an e-commerce environment, for example, then safety might not be so much of a primary concern, but we work in the automotive industry, where safety is paramount. We need to be certain that the algorithms that we’re putting into cars are robust, interpretable and explainable. In the automotive industry, if you want to bring AI into the car, you have to follow very rigid standards. We need to understand these and make it easy for developers to bring AI applications to life in the car.
A powerful AI-ready architecture
At CARIAD, we’re developing an AI architecture to serve all sorts of different AI use cases from various domains – whether that’s autonomous driving, mobility services, vehicle motion & energy, or intelligent cockpit & body.
This next-level architecture is we call the E3 2.0 architecture. It’s a single, scalable platform for vehicles across all segments, consisting of a customized compute power with AI acceleration in the vehicle, the capability to collect valuable data, and a powerful backend to efficiently and safely develop AI products.
We’re developing this platform to provide enough power and hardware capacity to support AI applications – and not just in the present. We can’t predict all the use cases that we’ll need to satisfy by 2030, but we need to make sure that the power and capacity is there to support future AI applications too.
In addition, our automotive cloud, VW.AC, provides a safe, scalable and holistic AI development toolchain. This toolchain can be used by any developer within CARIAD or the Volkswagen Group to work with the data, transform it, build a model, analyze it, and deploy it, all the while satisfying very rigid safety standards and regulations. By creating one scalable AI development toolchain, we can make optimum use of the scale of the Volkswagen Group and the synergies across the company. One toolchain makes it easy to exchange code and benefits collaboration.
The interplay between the AI architecture and the toolchain provided by VW.AC is going to form a platform to make it easy for developers to come up with fantastic new AI applications to drive customer value.
Intelligent AI applications and smarter future mobility
AI holds tremendous potential for the automotive industry and will be key to unlocking future profit pools for the Volkswagen Group. CARIAD is already investing heavily in this technology, and not just for the purpose of future applications – we’re already creating customer value with AI for customer today. Together with more than 120 data scientists working on more than 60 individual AI projects, our AI-ready E3 2.0 architecture will establish the basis for intelligent applications and lead to smarter future mobility.
Jan gave an in-depth talk about artificial intelligence at the IAA Mobility 2021 in Munich. To hear about the solutions from Jan himself, check out the re-stream of his presentation below.