Porsche, Highly Automated Driving and Artificial Intelligence
To drive, or not to drive, that is the question.
Dr. Sebastian Söhner, Specialist Data Driven Development and Innovation ADAS/HAD at Porsche, explains Porsche‘s hybrid view and plannings on autonomous driving and why Porsche will never get rid of their steering wheels as a statement for driving pleasure.
In recent years, autonomous driving has emerged as a key mobility trend. For some, self-driving vehicles are the future — a future in which drivers are freed from the burden of driving to make better use of their time while traveling.
At Porsche, on the other hand, we regard human driving as a privilege: Porsche stands for performance, perfect handling and pure driving pleasure. That’s why our customers want to drive their car themselves. As Porsche CFO Lutz Meschke recently said, “The Porsche sports car will be one of the last automobiles with a steering wheel.” If driving becomes less of a must, we see this as an opportunity to continue building the most pleasing and enjoyable sports cars in the world.
The opportunities of automated driving for Porsche
However, there are certain situations where driving pleasure gives way to frustration, and where automation could bring significant benefits to drivers. Let’s face it: no one likes to navigate through busy parking lots, and no one likes to be stuck in heavy traffic. Yet, such situations happen pretty frequently. In 2018, German drivers spent an average of 120 hours stuck in traffic. While we understand that our customers are extremely passionate about driving, we are also aware that highly automated driving holds tremendous potential.
For Porsche, automation is not a threat, but an opportunity — one that leads to completely new possibilities. In the following, I’m going to sketch some of these possibilities by focusing on Porsche’s approach to automated driving and artificial intelligence.
The Architecture of Assisted Driving Systems
Assisted driving systems are precursors to highly automated driving systems. The architecture on which assisted driving systems are based can be broken down into four basic components:
- perception,
- representation,
- planning and prediction,
- control.
Let me give you an example. Porsche InnoDrive is an assisted driving system that handles throttle inputs, gear selection and braking in response to conditions on the road. Using navigation data, InnoDrive calculates acceleration and deceleration values for the next couple of kilometers and modifies the speed of the car accordingly. If there are slower cars or speed signs, for example, the system decreases the speed of the vehicle. The driver is still responsible for steering, though.
But how does InnoDrive work? Looking at the four basic components, we have:
- Perception: Vision and radar sensors monitor the surroundings and extract relevant data from the environment, detecting and collecting real-time traffic data such as other vehicles or road signs.
- Representation: The system uses sensor fusion techniques to combine data from multiple sensors, thereby creating a comprehensive and robust model of the environment.
- Planning and Prediction: The situation is analyzed according to pre-established guidelines, and then local objectives are planned.
- Control: Based on the previous steps, the system predictively optimizes the vehicle’s speed.
Highly Automated Driving Systems: Data-management and Artificial Intelligence
At Porsche, we have developed a prototype for a remote parking pilot that allows the vehicle to self-drive and self-park in a parking lot — without a driver sitting behind the wheel. The driver leaves his car at the entrance, say of an airport, and then the car finds a parking space. In our scenario, the assigned parking space is equipped with a charging station. As soon as the driver returns from his holiday or business trip, he is picked up by a fully charged vehicle.
Highly automated driving systems are based on a comparable architecture as assisted driving systems. However, as you may imagine, the components along the event chain are much more complex and designed redundantly. To begin with, the driver is taken out of the loop, and we need a much more detailed mapping of the environment. To this end, we integrate a range of additional sensors into the car. In addition to these sensors, highly automated driving systems rely heavily on off-board solutions. Vehicles are connected with a backend and access high-definition maps. In fact, highly accurate positioning is required for these automated driving systems to work.
To deal with these challenges, a data-driven development approach process is needed, which provides us tools for validation and verification, as well as data-management processes. As a result of that, we establish an infrastructure to use artificial intelligence. But what role does artificial intelligence play here?
Bringing AI and Automotive closer together to leverage potential
Artificial Intelligence algorithms and machine learning have become exceptionally good at identifying objects, making highly automated driving systems more capable of processing traffic data and hence adaptable to continually changing scenarios.
We’re convinced that artificial intelligence holds enormous potential, but it has to fulfill certain automotive requirements in order to unfold its full potential. Meeting the challenges of highly automated driving also means being able to master AI in this area of application. For us, robustness and interpretability are the key performance indicators when using AI. Besides the application of AI for different use-cases, we have a closer look into the networks and try to understand why decisions were taken.
Concluding, I can say that highly automated driving systems are not going to change the essence of Porsche, but rather support to even further enhance the driving experience. With these systems, we will offer our customers an expanded range of opportunities — and we can’t wait to bring them on the road.
Dr. Sebastian Söhner is Specialist Data Driven Development and Innovation ADAS/HAD at Porsche AG. Follow us on Twitter (Porsche Digital Lab Berlin, Porsche Digital), Instagram (Porsche Digital Lab Berlin, Porsche Digital) and LinkedIn (Porsche Digital Lab Berlin, Porsche Digital) for more.
Are you interested in a job at Porsche? You can find all the details on careers and career opportunities at Porsche here. For a look behind the scenes, follow us on Instagram.
