Ways AI will Affect Automotive HMI
A recent blog post Nvidia invites us to imagine the future of intelligent vehicle interiors in a way that it builds trust through human-machine interfaces and with artificial intelligence.
The company argues that this can be done by offering personalized experiences, which are created by the interior design of connected and autonomous vehicles (CAVs) “for the next generation of transportation, including detailed visualizations of the car’s AI,” explains Brian Caulfield in Nvidia’s blog.
Andreas Binner, chief technology officer of Rightware, says there are two primary areas where his firm sees AI changing automotive interfaces. He believes that HMI software tools will add AI capabilities to “aid UI creators by automating basic user interface design and by optimizing the layout and configuration”. “For example, a designer might specify a cluster with a speedometer, a tachometer, a battery level, and range indicator, as well as an ADAS view. The HMI tool would then offer suggested designs and create a ‘pre-wired’ layout, ready for customization. Later, once AI-capable automotive computing hardware becomes more widely available, we’ll see digital cockpits which can reconfigure themselves based on user requests and actions.”
However, it’s not just about interiors when it comes to the application of AI. Stellantis revealed that the Jeep brand tested the latest prototypes of automated off-road driving technology with two electric off-road Jeep Grand Cherokees. The automaker wants to deploy AI to enhance the experiences of seasoned off-roaders, as well as of those who are new to driving off-road. It says the technology will also help Jeep SUV drivers in “challenging everyday driving conditions”.
Beyond UIs
Oleksandr Odukha, vice-president of delivery, mobility at Intellias says that he prefers to look beyond typical HMIs like screens or buttons. His preference is to concentrate on voice assistance in a car. He explains: “Today, possibilities to control vehicle functions by just saying the word are awfully limited, and the number of use cases could be much higher. Just look at ChtatGPT, which already illuminates industries like a Big Bang. Why don’t automakers adopt technologies like this?”
In his view the opportunities that voice control using AI and machine learning (ML) presents are huge. He says you can interact with AI, you can personalize and adjust its functions. The learning capabilities of the technology are key, including when it comes to the personalization of car settings for specific users. This means a car can learn an individual’s preferences and attach them to their profile, recognizing each person using a camera to adjust the car’s unique settings just for that individual.
“We are not only talking about your family users of your own car but also car sharing services that can provide AI-enabled cars capable of adapting to their users,” he suggests. In essence, he believes that AI and ML could impact automotive HMI that “we have not experienced yet”.
Need for human input
While there is a fear that AI, including about generative AI, could adversely affect society and humanity, Odukha doesn’t think that the technology will replace human input: “A human-being would still need to provide some input by voice or another interaction interface, while AI does everything else, even with such life-critical features as defining the driver’s health.”
AI won’t make human beings redundant just yet and it can in fact be used to work for us rather than make us redundant. He comments: “AI can recognize deviation from usual behavior when the driver falls asleep or feels seek and call 911. So, automotive HMI coupled with AI features will be a real differentiator for OEMs and a hot trend for years to come.”
Open AI standards
Niels Peter Skov Andersen, chair of the ETSI Intelligent Transport Systems Technical Committee says it encourages open standards as it wants to ensure there is free competition within the European market. “This is a way to avoid proprietary standard which in the end can lead to someone gaining a dominant position,” he says before citing the European Commission’s proposed a law on AI, the European Union AI Act of 2021.
The Act outlays regulatory requirements: “That need to be fulfilled by products using AI. When such new regulatory requirements arise, we typically experience that ETSI members are interested in developing open standards that align with the new regulation.” He then adds that it’s clear that AI will be part of future developments and that AI will be essential for the protection of vulnerable road users in traffic because they are often extremely difficult to detect.
For example, picture a 48-ton truck moving at 40-50mph. It’s relatively easy to have a good idea where it is and you can predict where it will be in a couple of seconds’ time. Yet, a school child is less predictable, as are road users with low kinetic energy. He, therefore, says: “It is in those cases that AI might play a role both in detection mechanism and prediction algorithms.”
Building strong partnerships
Binner explains that contextually aware AI can help drivers at the right time in the right situation. However, he believes there is a long way to go with automotive AI. The use of AI and ML is still nascent in the industry in terms of exploring their potential in HMI. To him, what’s exciting is the “previously unimagined functionality”, such as a car that recognizes a driver’s mood and changes a dashboard theme to reflect it, or the music selection to make that individual feel more relaxed.
Odukha thinks that much will fall down to the affordability of the technology. Much depends on the cost of implementing the AI and the computing power in the car, which could be relatively cheap. A low price point could see AI skyrocketing. He concludes by suggesting that if you offer a fair price and value for money, nobody will complain about the Terminator chatting from the backseat.
