4 pillars of Artificial Intelligence that are building the car of the future

Artificial Intelligence, autonomous vehicles, Deep Learning – these aren’t just buzzwords. The automotive industry is rapidly changing and in large part, it’s due to user demand. In these smarter, safer cars, it’s your interactions that will implicitly train the machine to learn your likes and dislikes – something made possible through these four pillars of AI.
By
Today’s connected cars are smarter with artificial intelligence technology

The automotive industry is rapidly changing. Autonomous and semi-autonomous driving are key trends everyone is talking about. They will not only change how we use cars, but—in the long term – how our (smart) cities are designed. Autonomous driving became possible by recent breakthroughs in Artificial Intelligence technologies—Deep Learning is one example.

There are other, very exciting ways Artificial Intelligence technologies help improve the driving experience. Case in point: Intelligent personal assistants. While most assistants proliferate on a smartphone, there is innovation happening behind the wheel that brings an automotive assistant to bear that fulfills your driving-related needs better, faster and safer. These are the four main pillars in AI being applied to the connected car:

 

Smart interaction

We make it possible for people to talk to and interact with machines – an experience that continues to become easier over time with advancements in Natural Language Understanding . In fact, being able to speak to your devices and cars is an expectation. You can say what you want in almost any way you like and have longer conversations. Now, the machines will remember what you said before, even if it was yesterday, and use that to get to the desired outcome more quickly and with less effort. Also, just like humans, the car will be able to recognize you by your voice through voice biometrics technology, which is useful to distinguish you from your spouse, for example.. That means even more personalization. So when you use your spouse’s car and say “Let’s drive to work,” the car will get you to your workplace, not your spouse’s workplace.

 

Contextualization

Since the car, by definition, is moving from place to place, the contexts in which it operates change frequently. You might be on the highway, or on a dirt road in the countryside. Maybe there is a traffic jam ahead, or a road closure. There may be sunshine at your current location but a snowstorm at your destination. Each of these can provide additional information that will make your interaction with the car more meaningful.

Context also includes sensor reading from the car itself, e.g. fuel level, crash avoidance sensors, or how many people are in the car. These parameters matter for making smart decisions when driving, and are taken into account by the contextual reasoning framework . Essentially, it’s artificial intelligence that thinks for you. The system knows that in a snowstorm, covered parking is certainly the preferred option. It will also try to find the cheapest parking for you based on your estimated arrival time and will find a gas station on your route that requires only a small detour, but is cheap at the same time. If you have a loyalty card, that will also be considered. It will even recognize that the traffic jam ahead will likely cause you to be late for your meeting, and offer to send your colleagues a note about your delay. These ideas aren’t simply flashy; our research shows that users are excited about cars that can make this a reality.

 

Personalization

Until fully automated vehicles arrive, decisions while driving can only demand a certain amount of attention from drivers, since the primary focus should always be on the road to enable a safe journey. But in many situations this is difficult to realize. Say you are looking for a parking space, or a restaurant for dinner, or you want to create a playlist with a certain type of music. In all of these situations there are too many options to choose from to make an informed decision while staying focused on the road. Enter personalization.

Here, you will start a dialog with your car where you can specify some core criteria of your request. This system will take everything it knows about you into account, and provide a recommendation, which you can accept or modify. Crucially, the system in your car will remember your choices, and take them into account the next time you are searching for something. You are implicitly training the machine to learn your likes and dislikes, and after a certain time many of its recommendation will become so good, that you can blindly accept them. And if you feel like something different today, you can always say so. This last part is key – NHTSA (National Highway Traffic Safety Administration) guidelines suggest that the driver should always remain in control of the conversation, and our own research shows that drivers want this ability even for systems they trust as very accurate.

 

Knowledge

Having the right information at the right time is also crucial when driving in your car. Consider two examples: Smart Car Manual and Travel Guide.  Smart Car Manual allows you to ask a wide range of information related to your car, such as, “How do I adjust the height of the steering wheel?” or “What is this yellow blinking light on my dashboard?” Especially when your car is new, or a rental, this functionality allows to quickly familiarize yourself with the vehicle. Another aspect of this is a smart travel guide: It allows you to ask questions about your current location, your destination, interesting sights along your route, or at your destination. You could ask, “What’s that building on the hill to the right?” or “What are the top sights at my destination?” Or simply, “What can I do around here?” is a great way to discover new things. It’s like your personal travel guide that is really knowledgeable about almost any place you visit, easily accessible while you’re on the road.

We’re at a crossroads and that’s what I found so exciting. What was once the stuff of fiction is becoming a reality. But, we’re not doing it for the sake of innovation; we’re building these intelligent car systems because even our user experience research says that drivers want it. Now excuse me while I get back to work to make this happen.

 

Tags: , , ,

Michael Kaisser

About Michael Kaisser

Michael Kaisser is Principal Product Manager for Artificial Intelligence technologies in Nuance’s automotive division, where he drives the company’s efforts to make its automotive solutions smarter. For the last decade he worked in various functions with a broad range on AI technologies, with a strong focus on Natural Language Processing and Machine Learning. He holds a PhD from the Institute for Language, Cognition and Computation at the University of Edinburgh. In the past Michael has worked at Microsoft Bing's Search Technology Centre as a Program Manager in the Core Algorithms group. As a Senior Researcher/Project Manager at AGT International he helped developing innovative Social Media Analytics methods and products in the Urban Management domain. Before joining Nuance he was co-founder of a Natural Language Processing consulting firm—txtData—in Berlin, where he advised companies on how to make their products smarter by using Natural Language Processing and Machine Learning techniques.