Sensors. Cameras. Radars. LiDARs. Computer chips… autonomous vehicles will require a multitude of these data gathering and management technologies to approach the current combination of mirrors, eyes, ears and brains. With so much data coming in and out – according to a recent estimate, one driverless vehicle will produce 4,000 gigabytes of data per day – these vehicles will essentially become moving data centers. Connected car data analytics, critical to both safety and initial viability, will play a key role in driving mass adoption and, when shared effectively, will produce a number of important benefits.
The public may feel safer if other drivers are compelled to keep their eyes on the road, but that may be an illusion. In fact, road safety is often cited by policymakers as a morally compelling reason to develop autonomous vehicle technologies.
The U.S. Department of Transportation’s National Motor Vehicle Crash Causation Survey (NMVCCS) of over 5,000 accidents over two years found that human error was by far the most important critical reason in the causation chain at ninety-four percent. Only two percent of crashes were attributed to vehicle failure. Another two percent was due to weather and road conditions. When drivers were assigned responsibility, recognition errors, decision errors and performance errors accounted for 41 percent, 33 percent and 11 percent, respectively. Therefore, in theory, fatalities will fall dramatically if we move toward less driver control.
Crash, near-crash and non-crash data will drive significant improvements in all vehicle design and driving protocols helping to ensure vehicles can safely turn into tight corners, navigate unpredictable junctions, deal with unexpected changes in weather conditions and road quality, and eliminate the all too common distracted driver errors.
Improved car design, lower emissions
Manufacturers already deploy onboard software that manages a vehicle’s engine use and which produces very granular technical performance data. This enables manufacturers to ensure vehicles comply with fuel efficiency regulations.
With autonomous vehicle technologies, the data collection and processing potential will multiply exponentially. Vehicles will literally become data centers.
We will be able to review all aspects of every type of driving event with reliable data as to pre-event movements, event data, causal factors and cause attribution(s), as well as all incidental and associated factors.
We will also be able to find optimum driver and software combinations and fuel efficient driving patterns (stopping distances, trailing gaps, etc.). We can observe which patterns result in more fuel efficiency and safety, and have empirical evidence to support changes in regulations for both manufacturers and road users.
Accurate fault attribution, lower premiums
Many law enforcement agencies already mandate dashboard cameras to assist with initiating prosecutions. The footage has helped prevent miscarriages of justice in many high profiles cases. While the power of video is clear, precise granular data can produce even greater benefits.
In May 2016, the first person to die in a semi-autonomous vehicle brought many moral accusations and grim company and industry valuation predictions. Tesla Autopilot was investigated by the National Highway Traffic Safety Administration (NHTSA) and the industry awaited its findings as to how the technology worked and it’s the role in the crash. In January 2017, the NHTSA concluded it was not the technology that caused the fatality and declared it unnecessary to investigate the technology in this incident any further. It seems the investigation was data-driven and very practical: the agency took a 2015 Tesla and a truck and conducted much-repeated scenario testing. The tragedy at least revealed the value of the increasing levels of computing power being applied to, and resulting data from, connected and autonomous vehicles.
It is quite clear now that a vehicle’s onboard data enables much greater verified understanding of what actually happens in those terrifying split-second instances. A combination of cameras, sensors and even voice recordings can help answer:
- What was the precise interplay between the driver (and software) and mechanical controls in those crucial moments?
- What opportunity did the driver (and software) realistically have in order to take averting decisions?
- What were the other road users doing before, during and after the crucial moments?
- What information corroborates or refutes various parties’ accounts of the event?
That type of verified data has simply not been available previously. For example, even a simple disputed insurance claim can drag on for months and sometimes years while expert evidence was compiled and challenged. Arbitrators have had to speculate on the reliability of third-party accounts of varying levels of detail.
Further, this more granular connected car data will, of course, help insurers better quantify collision risks and price that in terms of premiums that more accurately relate to the true level of risk being transferred.
Traffic management, infrastructure planning
Commercial fleet operators as well ride-sharing platforms already collect, manage and process data from hundreds of thousands of vehicles. They use this data to assess performance, to plan recruitment as well as geographical deployment, and determine peak pricing.
Transport regulators already use vehicle registration camera images to monitor traffic and manage its toll charging systems. Their informational potential is bound by the number of high-resolution cameras that can be economically installed.
Connecting car data will be critical
All of these in-vehicle and external data collection efforts will create many data lakes. However, they will be limited by being unconnected. Sharing the data will open up the full spectrum of data analysis potential. Data managers can compare third-party data with their own collected data for improved vehicle design, road planning and accident dispute verification and resolution.
Releasing the flow of driving data will create value for 4 main groups
As elsewhere in the modern economy, data is a critical raw material – and for autonomous cars and the advances in safety, public policy and commercial opportunities they’ll likely yield, it’s the key.
Coming soon, we’ll explore some of the legal and regulatory challenges around ownership and compliance of connected car data and analytics – watch this space.
In the race to put self-driving vehicles on the street by 2021, major automakers and suppliers are partnering with, investing in and acquiring
smaller component makers and technology start-ups so that they can expand their capability, pool expertise, share the workload, cut cost and get the cars to market sooner. View the special Reuters Graphics interactive report on the innovation boom in the automotive industry.
Thomson Reuters automotive solutions are here to guide you through uncertain times in managing risk and reducing cost in the supply chain.
Uncertainty and risk in the automotive industry – Download the report