Florian Leibert, co-founder and CEO, Mesosphere, says autonomous vehicles may be at the top of the hype cycle, yet many have yet to realise that the real story of transformation is not in the car itself, but the underlying technology.
Autonomous vehicles are moving ever closer to commercial viability. In 2016 only about 1% of vehicles sold were equipped with basic partial-autonomous-driving technology. Today, 8 out of the top 10 car makers have announced plans for highly autonomous technology to be ready for the road by 2025.
Cars are becoming computers on wheels – a complex mix of sensors, data pipelines, applications, on-board networking, user interfaces, and networked devices. Intel estimates that each autonomous vehicle will generate and consume roughly 4 terabytes of data for every eight hours of driving. Four terabytes is the amount of data it takes to store over 1.2 million photos, 70,000 hours of music, or 4,000 hours of video.
What’s more, much of that 4 terabytes will need to be captured, analysed and acted upon very quickly: at 65 mph, a 1 second latency could mean the difference between a car detecting and reacting to an object in time or being in a fatal accident.
Closer to the edge
It’s will be no easy task for 5G networks to transmit this volume of data from each car to the cloud and back. Just consider the staggering amount of data being generated and managed for tens of millions of self-driving vehicles in the coming years. To respond in near real-time, massive amounts of computing power must be available in close proximity to each car. Recent developments in edge computing, a decentralised extension of data centre networks and the cloud, bring computing closer to the data source and consumer.
Edge computing enables you to act on data quickly without the latency incurred by transmitting across a wide area network, which is essential in automotive scenarios. For an autonomous car, data that needs to be acted upon locally (such as an intruding vehicle) will be processed at the source.
The processing needs at the edge are quickly becoming more sophisticated, and even more so with developments in machine learning and deep learning – technologies that help to derive even greater insight from data. But there are also areas where latency is less of a concern, such as information about road and weather conditions, and in these cases the data can be pushed to the cloud.
This points to a need for better orchestration of the computing and communications between self-driving cars and other edge elements. Abstraction and automation could help enable this next wave of distributed computing. To manage a distributed environment from the edge to data centre to hybrid cloud requires a solution that solves a whole host of issues, including having the right team in place, overcoming regulatory challenges and mitigating the risk of permanent data lock-in to one place. Thankfully, the edge can be incorporated as an extension of the hybrid cloud design, creating a massively distributed and highly scalable computer system.
Autonomous car technologies are transforming the automotive industry with implications for automakers, ecosystem partners, and customers alike. They will hugely impact how computing and communications systems are created, both through hardware and software. And it is these advances in data management, combined with edge computing, that will help realise the full potential of cars, homes and cities of the future.
The author of this blog is Florian Leibert, co-founder and CEO, Mesosphere