Autonomous Vehicles Are Pushing the Boundaries of the Edge

Autonomous vehicles may be driving on Texas roads soon. The Texas Department of Transportation just announced its plan to create a Connected and Autonomous Vehicle (CAV) Task Force to become a central point for CAV advancement. The CAV Task Force will build on 2017 legislation passed allowing connected and autonomous vehicles to operate on public roads by enabling companies to pursue innovative ideas around CAV technology in a business-friendly way.

Texas is not the only state encouraging the development of autonomous vehicles. In 2018, Arizona opened the Institute of Automated Mobility to facilitate research and testing, and California and Virginia have also created testing facilities. These states are interested in the progress of autonomous vehicles as they have the potential to reduce the number of accidents and improve roadway safety over time.

Autonomous vehicles also provide opportunities to reimagine personal and commercial mobility with quality of life and economic benefits. For example, CAV technology could enable greater mobility for those who rely on transportation from others to access health care, such as the elderly and people with disabilities. There are also plans for subscription services that would enable consumers to summon autonomous cars on demand.

For all their promise, autonomous vehicles present enormous computing challenges. Each vehicle generates 1TB to 5TB of raw data per hour, primarily image files captured by the car’s high-definition cameras and LIDAR scanners. The car needs to understand what’s going on around it, so a large portion of the data is processed by the vehicle’s onboard computers. Because the latency associated with sending the data to data center is too great for real-time decision-making, autonomous vehicles typically carry 2kW to 5kW of compute power.

But not all data needs to be processed in real time. Some of it is saved for mapping, routing and preventive maintenance functions, as well as for training the machine learning algorithms that drive the vehicle. Onboard processing drains power, so it makes sense to send this data to a data center for analysis. The question is how to transmit the data when wireless connectivity isn’t universally available or cost-effective.

Most autonomous vehicles are electric, so they must be charged regularly. This affords an opportunity for the vehicle to download data while it’s plugged in and charging — just as aircraft engine data is downloaded when a plane is at an airport gate or in a hangar. And rather than sending this data to a centralized data center or the cloud, it makes sense to process and analyze it locally, in an edge data center.

Logistically, it’s not that complex. CAV development is primarily focused on fleet use cases, such as freight transport, local delivery and public transportation. These fleet vehicles will have to be charged in large facilities, providing a commonsense location for edge data center capabilities.

Building out an edge data center in a fleet facility comes with its own set of challenges, however. Operators will need modular, self-contained infrastructure that incorporates cooling and power management. Because these facilities will not be staffed with IT personnel, remote monitoring and management and high levels of automation will be critical.

Demand for edge data centers has grown alongside the Internet of Things and the need to process large volumes of data with minimal latency. Autonomous vehicles are simply pushing the boundaries of the edge with onboard computing and localized processing in fleet facilities.