Why Autonomous Cars Need High-Performance Storage

In June 1995, Popular Science Magazine published an article announcing the development of self-driving trucks by the United States Army. Led by the Tank Automotive Command partnered with a robotics company based in Pittsburgh. The self-driven Humvees were powered by an artificial neural network named ALVINN (An Autonomous Land Vehicle in a Neural Network). Much of how ALVINN was trained to drive required it to be in “learn mode” while a human drove it around. Subsequently, from the early 2000s until even now, the majority of research and innovation into autonomous vehicles derives from the military and industrial sectors.

The idea of autonomous cars began in the 1920s with a 1926 Chandler controlled by remote control by a person in a follow car. Despite decades of brainstorming, experimentation, and implementation, data storage has only become a topic within the last few years. In technology development terms, we’re living in the wild west. That means it’s almost pointless to guess the amount of data autonomous vehicles will be generating five years from now. It’s safe to say, however, that high-performance storage (HPS) will always be at the center of it all.

Self-Driving Cars and Industry 4.0

In the not-so-distant future, we and the world we live in will have begun reaping the benefits of an expanding global IT 4.0 DataSphere. According to Seagate, the global DataSphere is expected to grow exponentially within the next five years—175 zettabytes by 2025. In addition to that, approximately 30 percent of the world’s data will be processed in real-time. Data at the edge is crucial to achieving such an IT 4.0 DataSphere. Not only that but edge computing will be necessary for enhanced cybersecurity, artificial intelligence, and data creation.

Already, driver-controlled automobiles produce a lot of data with their sensors, infotainment systems, blind-spot vision, automated braking, collision warnings, and even cruise control. For example, late model automobiles feature 25 to 50 central processing units (CPUs) in order to manage the various technologies. In most cases, these vehicles host on-board networks that allow various systems to communicate with one another.

Gigabytes to Terabytes: Entering an Era of Automation, AI, Edge Computing

Autonomous and advanced driver assistance systems (ADAS) of a typical driverless car possibly produce close to 4,000 gigabytes per day—and this is but the tip of the iceberg. As of now, there still aren’t many people or companies using autonomous vehicles; however, what we’re seeing in data consumption is a portent of what is to come when wider adoption does take place.

But it’s not merely about how much data autonomous vehicles will consume; there is also the critical issue of managing latency. When an autonomous vehicle travels at 30 miles (48.28 km) per hour (48.28 km/h) or more, latency could very well become a matter of life or death. Employing the correct storage solutions allows onboard CPUs to process data lightning fast—fast enough for vehicles to respond in near real-time. However, that kind of responsiveness demands enormous amounts of computing power available, especially when these vehicles are near to one another. According to The World Economic Forum, IHS Automotive released a report predicting 21 million self-driving vehicles on the road in 15 years. Adding to that staggering number, 5G technology will be transmitting data from the vehicles to the cloud and back nonstop.

With the help of AI and machine learning (ML), what seemed impossible only a decade or so ago will be our reality tomorrow. And as IoT devices and automated vehicles become more dependent on unstructured data, cornucopias data points will supersede traditional datacenters. Even today, billions of devices around the globe are gradually bringing shape to our evolving DataSphere.

The Xanadu X-AI Series: Your Turnkey AI Compute and Storage Solution

AI and high-performance data storage are accelerating the progression of autonomous vehicles—vehicles capable of making AI-based decisions. For this, AI and ML solutions require storage that is optimized for huge quantities of small data transfers, data transformations, metadata, replication, ingest, and training. Our flexible and scalable Xanadu X-AI Series offers you a choice between the X5-AI, a hybrid storage array, or the NVMe X2-AI/X4-AI, an all-flash storage array. With the X-AI, your organization can scale-out with few TBs all the way to tens of TBs.

The Xanadu X-AI Series is a fully integrated, turnkey solution capable of scaling out capacity while maintaining superior performance. Designed with AI, ML, and DL in mind, the X-A1 handles flexible capacity expansion while providing 360 terabytes of scale-out NVM I400 appliances or 5.4PB of hybrid storage in the X500-AI. To learn more about Xanadu X-AI Series contact us today to learn more.