As cars become increasingly intelligent, manufacturers are presented with the chance to gain access to a huge amount of useful data. Harnessing this information to improve production processes will be key to the future success of British industry, says Jaguar Land Rover.
“If British manufacturing is to survive it needs to be competitive and it cannot be competitive without data,” said Adam Grzywaczewski, research strategy engineer for the Self Learning Car project at Jaguar Land Rover, speaking at the opening of Hadoop specialist Hortonworks’ new office in London yesterday.
Grzywaczewski explained the company’s vehicles are generating an increasingly amount of data.
“Our average car, on a good day, would generate about one and a half gigabytes of information,” he said.
People know modern cars are not purely mechanical, he continued, “but it comes as a surprise to many that we have currently about 60 computers onboard”, as well as fibre optic networks and "thousands of sensors".
The growth of data is a common theme across the manufacturing industry. Speaking to ComputerworldUK, Virgin Atlantic IT director David Bulman previously said its business is preparing for a significant increase in data as it embraces the 'Internet of Things', with a new fleet of highly connected Boeing 787 planes each expected to create over half a terabyte of data per flight.
JLR's Grzywaczewski said by harnessing data that can be collated from a variety of sources, vehicle manufacturers will be able to optimise production processes. This will allow firms to become more cost efficient, while also improving quality and safety of its products.
For example, many cars are 'over-tested' due a lack of knowledge of what a driver's typical use of their vehicle is.
“Currently cars are designed based on worst-case scenarios, with very well-defined tests. But that needs to be more informed by understanding how people use our vehicles,” said Grzywaczewski.
“We very frequently significantly over-engineer our cars because it is really difficult for us to tell what the normal use case is for normal vehicle components – what is the normal usage for disc brakes? We can estimate, so we will design against millions of events and we will literally test millions of events in various combinations, but to be honest we don’t know."
He added this decision-making process will be boosted by the growing use of telematics, which will offer manufacturers data on a "much, much wider scale" in the future.
“So hopefully what we are doing now will allow us to make much more appealing products like we are doing with the self-learning car, but also drive down the unit cost and increase quality. Some of the stuff we do around prognostics and interactive service is part of that future.”
Adopting new technologies has been key to making sense of the data available to the firm. For JLR, this has meant investing in Hortonworks' open source Hadoop database technology.
Grzywaczewski said: "JLR is two different brands so we have inherited vehicle architecture from both parents of ours - we we have a lot of data and a lot of data variance across different models. Even though everything is very well documented, there isn’t a very straightforward schema with which it could work.
"We couldn’t just go for a relational database and relational analytics, we had to think outside of the box. Obviously the capability to quickly validate our algorithms on a daily basis was essential to us."