There is a supercharged race to develop connected car technology and beyond that autonomous cars – street safe, legal, self-driving automobiles.
Traditional car manufacturers, including General Motors, Ford, Daimler, Toyota and Honda, are both cooperating and competing with tech giants such as Apple and Google.
Then there are the upstarts, including Uber, Lyft and Tesla, which are fundamentally disrupting the industry, creating new markets, new business models and driving new technologies.
The competition is producing fundamentally different approaches. Ford, for example, is charging ahead with a fully self-driving car concept, while Tesla, known for its engineering excellence, is going for a more incremental approach. Tesla founder Elon Musk, meanwhile, has criticised Google’s use of Lidar (Light Detection and Ranging) technology as unnecessarily complex. He has equipped his firm’s autonomous driving test vehicles with passive optical sensors and a radar system.
And competition isn’t only enveloping the automotive and technology industries. Governments are getting in on the act, too.
In his final budget, then Chancellor George Osborne said the government would establish the UK as “a global centre for excellence in connected and autonomous vehicles”, and promised to:
- Conduct trials of driverless cars on the strategic road network by 2017.
- Consult this summer on sweeping away regulatory barriers within this parliament to enable autonomous vehicles on England’s major roads.
- Establish a £15 million ‘connected corridor’ from London to Dover to enable vehicles to communicate wirelessly with infrastructure and, potentially, other vehicles.
Whether this survives the post-Brexit referendum change of government remains to be seen, but Osborne’s ambitions already lagged behind those of Germany, which has put digital at the heart of its transport policy by creating the Federal Ministry of Transport and Digital Infrastructure. In an uncompromising policy document, Strategy for Automated and Connected Driving, the ministry declares:
“The motor car was invented in Germany… All the major innovations associated with the car – from the four-stroke engine to the anti-lock braking system – come from Germany…. Now digitalisation is about to usher in a historic revolution in the field of mobility – automated and connected driving,” and “we want Germany to determine the digital innovation cycle”.
The UK and Germany are, of course, not alone in declaring their intention to be world leaders in autonomous car technology. Similar statements of intent can be found from governments around the globe.
The stakes could hardly be higher for auto manufacturers, tech giants and nation states, and neither could the technical challenges – most of them centred on data.
Connected cars are already creating a data explosion. Analyst group Gartner says, “The production of new automobiles equipped with data connectivity, either through a built-in communications module or by a tether to a mobile device, is forecast to reach 12.4 million in 2016 and increase to 61 million in 2020.”
The connections made by these units include vehicle-to-vehicle, for example in adaptive cruise control; vehicle-to-infrastructure, particularly in Smart City technology; and vehicle-to-manufacturer.
The range of benefits these connections can deliver include improved safety, reduced congestion for the citizen and proactive vehicle maintenance, alongside design insights and software-based updates and upgrades to the vehicle for the manufacturer.
The data created by the connected car revolution will challenge us in terms of the sheer volumes created and in its effective capture, management, security and storage. Above all there is the challenge of turning this data into actionable insights – to deliver the evidence base for real-time action.
However, the volumes of data generated by connected cars are insignificant when compared to those generated by autonomous, self-driving vehicles.
The challenges include:
- The complexity of the tests required to deliver proof of concept for autonomous driving.
- The volume of data produced, which can amount to petabytes.
- Data management and transfer problems at test locations, which are often remote.
Traditional road-testing techniques, where data from trials is collected and sent back to headquarters to be formatted, sorted and analysed, are simply not viable for the huge volumes of data created by autonomous car trials or the speed at which the technology is planned to come to market.
Current test data volumes are in the multi-terabyte range and data pipes cannot transfer these quickly and efficiently. As a result, the information stored in databases is usually physically moved by the testing engineers, with a delay of days or even weeks before it is available for analysis.
To enable autonomous driving proof-of-concept tests to take place, Hewlett Packard Enterprise (HPE) and its partner NorCom Information Technology developed a new approach, so that data can be analysed promptly and necessary changes to test vehicles’ algorithms can be performed quickly to enable new tests.
They took the concept of the data centre in a box and enhanced it to create an analytics in a container service, which can be deployed on location for road tests or on suppliers’ premises.
Matthias Bauhammer, Practice Principal for Analytics and Data Management at HPE, says, “What is special about this mobile solution is that we combine a big data analytics platform with a powerful big data server architecture, where all components are optimised and geared towards fast data analysis.”
With the hardware and software containerized and moved to the location where data is collected, for example to suppliers’ facilities or to test tracks, Dr. Tobias Abthoff of NorCom, says, “Analyses are performed on location, even though the responsible data specialists are seated in their offices in the development centre. Instead of measurement data, simply analytic commands and analytic results are being transferred.”
If the results suggest calibration or algorithm adjustments, these can be implemented quickly, and therefore applied with minimal delay in further test drives.
According to Bauhammer, this approach doesn’t just deliver speed of analysis and help manage and control the vast floods of data that are being generated in tests.
“Current architectures will not be able to handle analyses of multi-terabyte or exabyte data volumes,” he warns. “Big data architectures will be required, allowing for an analysis and automatic classification of these data volumes. This makes for efficient decisions regarding which data needs to be stored, which information is necessary for analysis and which can be erased.”
While the drive for autonomous cars is producing mountains of data and new methods of handling it, the analytics in a box approach is not only suitable for the automotive industry. Aviation, other manufacturing industries and the energy sector could all make use of these new approaches.
This article is brought to you in association with The Business Value Exchange