Continental’s acquisition of Finnish automotive specialist Elektrobit for 600 Million € just made headlines in all major news channels. As EB is especially strong in HMI, this step has major implications for the relevance of Human Machine Interface for the automotive sector. It is hence appropriate to take a look at the future of automotive HMI.
The two broader issues concerning both OEMs and suppliers are mobility and connectivity. Mobility here is not only connected to driving. We need to consider new forms of using vehicles as such, in order to channel HMI-systems accordingly right now. Future mobility will in all likelihood develop in a continuum of shared mobility and autonomous driving. US streets are already home to a few dozen autonomous cars, a trend set and driven especially by Google. Apart from Google’s prototype, there are Toyota Prius, Lexus RX450h und Audi TT.
Market analysts and specialists expect a commercial use of the field research in autonomous driving in about 8-10 years. Jaguar/Landrover expect their autonomous cars as early as 2024, Audi even in 2023. After all, German Federal Minister of Transport Alexander Dobrindt recently rode a semi-autonomous Audi a7 named Jack along a section of the A9 which had been developed as a test section for autonomous cars, and journalists have travelled from Silicon Valley to CES Las Vegas – a distance of 900 kilometers which the Audi was able to complete within two legs.
The technology for autonomous driving is quite advanced already, and, considering the development cycles in the automotive industry, this is the right moment to customize the user interfaces and infotainment systems of the near future according to these requirements and terms. Relieved from the cognitive load to constantly focus on the traffic at all times, „passengers” are able to explore new ways of interacting with their vehicle. Robert Zetsche, CEO at Daimler, sees a shift in priorities in relation to the relevance of the car of the future. The perception of the car will change from status symbol to that of a private space allowing passengers time to tend to professional and private matters.
Privacy and quality time have already advanced to become the most important luxuries today, and their value will continue to increase as the world is spinning faster than ever. Multimodal control options and assistance functionalities in the car enhance the value of these luxuries immensely, and intelligent and natural language interfaces will soon become an integral part of the connected car.
Shared Mobility enhances the demands placed on the connectivity of such interfaces. Thousands of drivers are already using options like DriveNow, car2go, Flinkster, multicity and so on. This makes perfect sense, as a recent study by Morgan Stanley reveals that private cars sit idle for 23 hours a day. The resource and investment into the private car thus appears to be quite inefficient, and the use of the car by several people bears the potential to solve this discrepancy, not to mention acquisition costs. Shared Mobility also entails, however, that one driver might have to use several different cars, and that different drivers might use one and the same car. Those who do not wish to forego their valued freedom in communication and control when using Car Sharing, will rely on safety and personalization concepts which the development of new user interfaces will have to deliver. Standards play an important role, as do flexible platforms like SemVox ODP S3, which may function as an aggregator for all types of functions and modalities and are powerful enough to implement any conceivable assistance and control function.
Contextualization is another buzz word: which content is displayed at a particular time, to a particular user, and for a particular reason? Genuine, AI-based language comprehension and semantic routing will have to differentiate „Big Knowledge“ from Big Data. Rendering unstructured data searchable is a necessary precondition for search options but it is the processing of information to filter and recognize preferences, e.g., personalized search setting on a general or situation-specific level, which makes an assistance system intelligent.
The car is definitely geared towards the Cloud, with an integrated SIM-card or via a smartphone connection. Hybrid systems are gaining importance as teething troubles such as the latency times in cloud-based speech recognizers and NLU-systems are fewer and because the robustness and performance of speech recognizers which are necessary for the personalization and generalization can to date only be delivered by cloud-based recognizers. The right balance between the situational use of Embedded and Cloud is of importance here: the vehicle‘s Head Unit is responsible for the suppression of ambient noise, and for specific domains there are robust voice recognizers which do not depend on an internet connection. The speech synthesis is being performed locally, i.e., embedded, which makes it independent from the Cloud. However, what is happening in the Cloud? Well, any kind of interaction with web-based apps will take place there. Facebook, streaming services, chats and so on need an internet connection, and specific requests (e.g., ticket orders or table reservations) rely on being connected to the internet. Cloud-based speech models can be updated easier and quicker and offer additional features, which cannot (yet) be catered for by embedded systems, e.g. dictation.
The amount of data processed and analyzed in the Cloud (e.g. error- and trend analyses) and the update capability of cloud-based services, in the interplay with local components, is an important factor which, together with the local components, contributes to making intelligent and complex functions more convenient and comfortable.
Trends in AUTOMOTIVE HMI:
- Voice-Control and Dialogsystems with genuine multimodality, which meet the demands of contemporary mobility and connectivity through
- Personalization and Contextualization of Assistance systems, using AI-technologies and …
- hybrid, customized processing of language