Software-defined vehicles are dramatically expanding the possibilities for automotive development, testing, services, updates and business models.
The building blocks are all here — an ever-growing number of sensors, increasingly flexible onboard computing platforms, widely available mobile networks and cloud-based storage and analytics. This is a pivotal moment for new automotive applications, and vehicle manufacturers, suppliers and others invested in the future of mobility recognize that it will be software-defined.
Data is key to these advances, but most of today's tools for collecting and using data generated by vehicles barely scratch the surface of its true potential. The automotive industry needs new technologies to make emerging data-driven applications both practical and economically viable.
Most vehicle data solutions are constrained by the attributes of traditional automotive engineering, designed around purpose-built electronic control units developed for dedicated functions that remain largely unchanged for the life of a vehicle. This same approach has been employed, in most cases, for data collection. In these fixed approaches to data, automakers are usually constrained to collecting certain limited types of bulk data and uploading it over the air. Some third-party users of vehicle data, such as insurers, typically access data streams through added in-vehicle hardware. This adds cost, requires user involvement for installation, and lacks tight integration or scalability.
Now, time-honored automotive development practices are giving way to new approaches growing out of data centers and the cloud. Key principles of data center architecture include the ability to constantly upgrade software and the rapid evolution of analytics tools. Using these principles, automakers are now able to free software and services from the limitations of existing approaches where software was largely fixed.
Bringing a data center mindset to vehicles introduces powerful capabilities throughout automotive technology and business models. It enables multiple simultaneous data streams to internal and external users, including suppliers, insurers and fleet operators. It lets automakers and clients collect any data of interest that the owner allows. A cloud-based platform can define what types of data to upload, from which vehicles, how frequently and when, and those settings can be updated almost immediately to vehicles — even to vehicles in motion. These features are necessary for a return on investment that transforms vehicle data into a sustainable business.
A core component of these breakthroughs is the ability to configure vehicle data streams through policies rather than software updates. While over-the-air updates give automakers greater flexibility to improve vehicles after the sale, it can take months to modify, test and distribute new code. By contrast, policy configurations are only a few kilobytes in size and can be developed without programming, then quickly validated and deployed in seconds, even by third-party clients, which significantly reduces cost, complexity and risk. Data capture and processing becomes a unified, accessible service offering almost any number of policy configurations.
An insurance company, for example, could configure a policy to collect data on drivers' braking to determine risk. Subject vehicles would capture and upload specific data points, such as rate of deceleration, just when the brakes are applied. If necessary, the policy can easily be updated to collect more data if the initial policy did not provide sufficient information. This cycle of learning and continuous refinement is powerful and in stark contrast to the typical long development times of conventional approaches.
A data capture model with fine-grained policies can tame an otherwise overwhelming flood of data from increasingly smart, connected vehicles.
For example, an automaker doing field diagnosis on a vehicle to avoid a recall can set a target to capture and send data only from affected vehicles, such as those with particular hardware, software versions or manufacturing dates. This significantly reduces demand on mobile networks, cloud connections, cloud storage and computing capacity. Not just the automaker, but customers, mobile carriers and cloud providers all benefit from a model in which high-value data is sent in a configurable way. Ultimately, this may also lead to a reduction in vehicle recalls and improvement in consumer satisfaction.
Scalable, affordable access to essential vehicle data is now within reach for the entire automotive ecosystem, which represents an exciting time for the industry.
