May of us probably take the cloud for granted – we’re used to being connected wherever we are: at home, on the road , at the most remote places of the world. 35% of new cars sold globally in 2015 offer connectivity and this number is expected to grow to close to 100% by 2020. We Rely on the cloud when we’re behind the wheel for all kinds of advanced features and up-to-date information, such as navigation, music, traffic and much more.
Find more statistics at Statista
Connectivity isn’t always an option
But the reality is that constant, ubiquitous connectivity is often wishful thinking. Connectivity in rural areas can be problematic, as well as on highways or in street canyons in large cities. Data networks are often congested.
Source: Computerworld mobile data service survey, 2015. Base: 870 respondents
The German IT magazine Connect recently tested data connectivity while driving in German, Austria and Switzerland. Depending on the operator and areas, data connectivity ranges between 25% and 90%.
As a result we see an increasing demand for automotive infotainment solutions that are less dependent on data connectivity. It shouldn’t be required to have a cloud connection, if all that I want to do is enter an address into the onboard navigation system in my car. Or to select my favorite music on the USB connected to the car. Or when checking for the oil status or tire pressure. Take messaging for instance. Trying to dictate a message can be frustrating when connectivity is low – or worse – distracting when you’re checking the screen to see if it’s gone through.
Another side effect of a completely cloud-based service can be the cost of connectivity. Obviously, accessing data in the cloud comes at its cost, but sending voice data over the data network for recognition purposes can create cost, too, and unnecessarily.
Ideally – you need both – embedded and cloud-based solutions. Automakers are looking for solutions that combine the best of the two worlds – depending on network availability, cost and user preferences.
An example of how this hybrid approach works is in the recent BMW vehicles that processes simple requests such as selecting a radio station or address entry embedded in the car, other tasks such as search for points of interest are processed in the cloud when connectivity is available, and otherwise in the car.
And soon, we’ll see cars that can process complex speech recognition tasks such as embedded dictation of free text all embedded. People can dictate messages, e-mails, notes and more in the car, even when data connectivity is unavailable or unstable, avoiding user frustration such as latency, interruptions or failures.
Embedded dictation also addresses consumers’ data privacy concerns for those who have them. By processing the recognition of the messages in the car, car manufacturers can provide an alternative to the cloud where consumers want it.
While in a perfect world the boundless power of the cloud can process any task at any time, for the foreseeable future we can expect limitations in network connectivity, resulting in user frustration. Tasks that can be performed via an embedded system should leverage the processing resources in the car. Gains in processing power of embedded platforms as well as performance gains resulting from new technologies such as machine learning, allow embedded platforms to perform a whole range of tasks.
But it’s a balance – and why hybrid platforms deliver the best experiences. Use cases that rely on dynamic data, such as up-to-date POI information, are today better processed in the cloud. Ideally, tasks processed in the cloud are enhanced with a robust and reliable fallback embedded in the car. So that next time, when I’m approaching my home, I can still finish dictating my messages without being interrupted when the connection gets poor.