Home > Analysis > It’s time to push connected cars right to the edge

It’s time to push connected cars right to the edge

Connectivity will enable everything from connected services to autonomous cars and maybe even better-selling EVs – but how will connected cars cope when off-network? Xavier Boucherat talks edge computing with Kal Mos of Mercedes-Benz

The number of connected vehicles on the road continues to proliferate, as consumers seek to take advantage of new convenience and entertainment features. Connectivity has become so engrained into the fabric of everyday life that before long, the unconnected car will become an economically unviable prospect for OEMs.

That’s according to Kal Mos, Vice President for Connected Car, User Interaction & Telematics at Mercedes-Benz Research & Development North America. “Our phones are connected,” he says. “We’re connected all the time. And so getting in the car and finding you’re not connected can be a stressful experience. As such, connectivity is already written into the DNA of our cars – there will be no cars without connectivity, period.”

Improvements to the underlying technology continue, along with innovation in the connected services sector. But looking ahead, Mos predicts that as demand for services grows, OEMs will be faced with new challenges, all of which must be immediately taken into account.

One of his primary concerns is a fundamental one – the availability of reliable connectivity outside of built-up urban areas. As OEMs work to create connected features, so too will they need to make sure that these features can function without connectivity.

Key to this will be edge computing, in which data-processing is decentralised from a central unit in a network and pushed to the periphery – in this case, the car. “If you do get connectivity, you get enhanced functionality and features,” says Mos, “but nobody wants to be stranded when not connected. People assume the car has everything they need, and they assume connectivity is guaranteed, when in reality it isn’t. This is why we place such importance on edge computing, and this is one of our biggest challenges.”

We’re connected all the time. And so getting in the car and finding you’re not connected can be a stressful experience. As such, connectivity is already written into the DNA of our cars – there will be no cars without connectivity, period

Edge computing makes use of different techniques which help features work without a connection. For example, there are several techniques that could help a music-streaming platform continue working outside of well-connected areas. Tracks could be pre-cached, for example. That said, Mos believes that for the time being, customer attitudes are such that most drivers will understand that some non-essential connected services will not be available in all areas.

Where failure will not be tolerated is in critical systems, and this includes navigation systems. “If you are navigating and waiting for your right turn, but suddenly you lose connectivity and directions, that could be a real problem,” he says. “Customers would not accept that.” Several mapping apps already have offline capability, allowing users to download map data and relying on technology such as GPS to track movement.

Development of edge computing capability will also prove essential in enabling artificial intelligence (AI) within vehicles, which among other things will aid in the development of the autonomous car. Hugely powerful processors, capable of crunching massive amounts of data, mean that AI  capabilities have increased tremendously in recent years. In certain applications, such as static image recognition, some can already outperform humans.

The problem, says Mos, is that work to date has been primarily carried out on server farms, where connectivity is largely guaranteed. The challenge lies in bringing that power to the car, where resources are more limited. “Where do you store all the information needed?” Mos muses. “How do you process it, and what kind of power will this require to make real-time decisions? Eventually we will have to reach a point where we enable AI and machine learning in the car and get results in real time.”

In addition, says Mos, it is important to understand just how much greater a challenge driving presents when compared with something like static image recognition. Images continually change when driving, many more variables are introduced and different weather conditions mean that image quality can differ significantly.

“And even if we get this right,” he says, “something like image recognition is just one of several human capabilities. Drivers are not just watching – they are hearing, feeling, and calling on their years of experience around things like locations. They know where is safe, and where is crazy. They know where there are schools, and when children may be running about. They have a total knowledge.”

For these reasons, Mos suspects the move from lab AI to edge AI will take a huge amount of work, but with connectivity set to enable advancements across a number of priorities, the value for OEMs is plain to see.

Connected, automated, shared, electrified

Along with accelerating developments in the field of autonomous driving, connectivity will spear advancements in shared mobility services and electrification. “We will reach a point where connectivity will enable many things behind the scenes,” says Mos, “all of which will make a customer’s experience much more fruitful.”

Nobody wants to be stranded when not connected. People assume the car has everything they need, and they assume connectivity is guaranteed, when in reality it isn’t. This is why we place such importance on edge computing

Electrification in particular remains a priority for many OEMs, as infrastructure improves and emissions standards tighten worldwide. In the future, says Mos, Daimler expects that connectivity will enable complete visibility of an electrified vehicle’s status, including detailed information on a battery’s charge and condition that will help a driver to make better informed decisions to keep the vehicle running. Furthermore, connectivity can help the vehicle better communicate its needs to charging infrastructure.

“For example,” he says, “when a vehicle returns home and parks, it should be able to negotiate with a home charger how much time it needs, as this can limit costs. The same thing could be offered at the office.”

In addition, connectivity will help OEMs like Daimler build better EVs and hybrids by monitoring their performance and the behaviour of their drivers in different environments. “For example,” asks Mos, “are they charging every day, and for how long? Are they driving 50 miles a day, or is it more like 500? And beyond that, we want to understand different customer segments across different demographics and markets, with different lines of car available. We can understand all of this through connectivity, and we can make better cars as a result.”

Mos stresses that, just as with many other connected services, it will be important for customers to fully consent to the collection of such data. Privacy remains an extremely important issue, particularly for a German company like Daimler which is required to clearly outline what customer data is kept, where it is kept and how it is used. All connected services offered by Daimler, says Mos, begin with the customer understanding and accepting.

However, Mos also suspects that attitudes among first-time electric vehicle users may differ from those of internal combustion engine users. “When you buy an electric car,” he concludes, “the experience of driving completely changes. In a gasoline car, it is possible to stop at any station and quickly fill up. But with EVs, drivers start having to think about when, where and how to charge.” Mobile apps will play an important role, he adds, giving drivers continuous access to a vehicle’s status.