More than just autonomous driving: artificial intelligence in the automotive industry
AI plays a key role, both in the development of new types of vehicles and also in production – and is becoming a competitive factor. Continental is planning a smooth transition from highly developed driver assistance systems to the automated car. However, AI facilitates more than autonomous driving for the automotive industry. One focus for the use of AI at Continental is currently the vehicle interior. It should offer more than just comfort. The functionality of the interior, which interacts with its passengers in various ways, decides whether or not automation will be accepted. Hardware is thus taking a step back – while software is taking center stage. “For our development, the cockpit of tomorrow means an increasingly stricter separation of software and hardware as well as consistent openness to open-source software from many sources and therefore also to new business models.” says Michels.
There is a particular focus on communication between the user and the vehicle. AI creates the basis for an intuitive dialog between people and machines. To this end, it must be able to recognize individual drivers and interact with them appropriately.
One example of this is that each driver has a different driving style. What one user perceives to be too fast is considered to be too slow by another. Braking, overtaking, acceleration – all of these behaviors are specific to the individual. If automation takes over driving in the future, it may indeed navigate without any accidents but nonetheless annoy the user with incorrect driving behavior. The machine driving characteristics only need to deviate minimally from those of the driver for this to happen. It comes down to fine details.
AI learns appropriate behavior through algorithms and empirical values. “The vehicle must guide the passenger without patronizing so as to offer the required functions from the vast range of possibilities in an understandable form,” says Michels. It is only when the machine consistently integrates observations of the driver into its own behavior that user satisfaction increases and thus also acceptance of self-driving vehicles. The interior of the vehicle must be capable of empathy, of recognizing the user’s feelings and perception and of converting this into its own appropriate actions. Continental uses different kinds of sensors for this. Besides voice and face recognition sensors, these also include new types of medical sensors that measure heart rate, blood sugar and blood pressure. AI already recognizes the driver’s line of vision and degree of attentiveness.
Continental already provides display and operator solutions such as 3D touch displays or head-up displays that use a lot of this information. Thanks to HPC, the borders between various screens in the modern cockpit will be removed. The systems also help to improve the safety of the occupants by detecting tiredness and line of vision. “We are thus raising human-machine interaction to a completely new level and laying the foundation for intuitive communication in the networked cockpit of tomorrow,” says Frank Rabe, head of Human Machine Interface. And development is making strides. Robert Thiel, head of AI in Advanced driver Assistance Systems, promises: “We will soon launch new AI-based products. And we still have space in the data center.”
In order for this to succeed, Continental uses raw data as the starting point for a process which continuously improves the AI functions, including in the interior. Besides the internal data, external information is also used to improve AI. This includes environment data (temperature, humidity or brightness) or data from the automotive company (such as the availability of certain spare parts). The empathetic interior meanwhile serves as the main information source for allowing autonomous driving to become reality as standard. And the data remains in the vehicle. Only anonymized information is sent to the cloud for AI optimization.
In production, AI supports robots and machines in improving how they work. Software plays an important role in the automation of manufacturing. Machines even learn from mistakes and automatically optimize their behavior. That is because: “When machines are trained by humans, we can only teach them what we know ourselves. With machine learning, they learn to be ahead of us,” explains Balázs Lóránd, head of the AI center of expertise in Budapest. The method creates an artificial neural network between the machines, which helps to learn from experience and to combine new information with existing knowledge.
Questions regarding how AI can be reconciled with the ethical responsibility of the company toward its customers are posed more and more frequently. “For our products, services and work processes, there are 365 data protection days in the year,” is the motto at Continental. Data protection is not an optional extra. In the context of AI, the Group has formulated a code of ethics.
The maxim is as follows: AI that is developed in accordance with ethical considerations should support and relieve the burden on humans. The guidelines comprise eight points and are based not only on all of the legal requirements of the countries in which Continental is active, but also on the EU ethics guidelines. Maria Anhalt, CEO of Elektrobit, says: “As AI and the associated values and ethics are being constantly further developed, we will also adjust our ethical guidelines.”
Back to software overview.
