Table of Contents
Self driving cars are turning dreams of the future into a modern reality, and as the technology matures, public and private transportation will change forever. After all, cars destroy people’s driving, and drowsy, dangerous, incompetent, and distracting drivers disappear from the road. According to the National Highway Traffic Safety Agency (NHTSA), a single-car accident in the United States in 2017 killed less than 40,000 people, about 90 percent of which were caused by people.
But what is behind this technology? Why is a cars safer? What do you need during the day to travel without paying attention to the road?
Artificial intelligence drives self driving cars
To automate your car, you need to keep an eye on your surroundings. First, it recognizes (identifies and classifies) the data by driving the vehicle automatically / computer; then it acts according to that data. Self driving cars must have a safe and fast solution that allows them to make immediate decisions based on a detailed understanding of the driving environment. To understand the driving environment; various sensors in a vehicle must collect large amounts of data and process them in an autonomous vehicle computer system.
Artificial intelligence (seeing, understanding, and making the right decisions in all potential transport conditions) is required to drive an artificial intelligence (AI) vehicle without user control. Fully formed the computational performance of self driving cars, can be compared to the computational performance of some well-functioning platforms that were possible a few years ago.
Cars are expect to have more code than any other software platform. By 2020, model vehicles are expect to have more than 300 million lines of code and more than 1 terabyte (terabytes) of storage space. Also, a memory bandwidth of more than 1 TB per second is require to achieve the computational performance require for stand-alone platforms.
Self driving cars artificial intelligence systems make real-time decisions based on complex data that requires a continuous and uninterrupted flow of data and instructions. While there are successful examples of self- driving cars today; many of these early car crashes use the same road for several days, learning every detail and creating high-resolution maps. This is the result that was created and use as an important part of the automated navigation system.
You can focus your automated computer’s attention on the traffic situation, pedestrians; and other hazards in real-time without too much belief in the need to know the road. This limited driving is commonly refer to as a profession and reflected the approach that the first autonomous vehicles introduced unmanned vehicles. Geolocation leads to efficient solutions on limited roads; but autonomous vehicles based on geographical location may be successful in some areas, but not in others.
Memory: Hero behind autonomous driving
Storage of data from hard disk drives (SSDs) NAND flashes, NOR flash, low-power DRAM, GDDR6; memory subsystems associated with touch synthesis, routing systems, or black box data systems. Storage devices In the future, this will allow you to respond to emails; answer Skype calls, and watch your favorite shows when your email is in a safe direction to reach your address.
According to the chief executive of automotive systems engineering at Micron’s compact business unit; high-quality AI-based computers allow self driving cars to move better than human cars. Uses a deep neural algorithm.
“There are a lot of different sensors that seem longer and more accurate than humans; and they work together to see the environment 365 days a year, 24 hours a day, 360 degrees,”
“In addition to the maximum computer skills that car owners can have today, situations can allow us to move more safely than people.”
Imagine a car suddenly braking on a noisy road. By adding vehicle (V2V) / vehicle (V2I) communication (called V2X); this event will be transmit wirelessly to all subsequent vehicles, allowing each vehicle to inform the situation in flight. You can prevent accidents by grasping, actively braking, and braking.
High-speed memory is an essential part of self driving cars
Do you remember the stats according to which about 90% of the road accident victims in the US in 2017 were due to human error? When faced with unexpected risks, people can easily be distract and make quick decisions. On the other hand, computers are not distract and can interact continuously in less time than human impulses.
Surprisingly, safety is a top priority for self driving cars. From a security point of view, it includes an appropriate infrastructure that allows communication between vehicles; Or the surrounding environment, thus preventing miscalculations between redundant hardware systems. The redundancy of the computing subsystem of wirelessly connected devices is subject to legislation to determine the required level of security; which is directly related to the automatic level.
Relationship between human and computer control
The National Road Safety Administration (NHSTA) has established a set of standards that define the relationship between human and computer control; And vehicle management to monitor the development and implementation of autonomous driving technology. These range from 0 degrees (no driving automation) to 1 level (driving assistance), 2 levels (partial driving automation; the driver must hold the handle), 3 levels (driving automation). Conditional driving depending on the driver and its levels. It can be divide into intervals. 4 (advanced driving automation) and finally 5 degrees (fully automatic driving). Most ADAS solutions on the market today are based on computers with relatively mature Layer 2 functionality and memory elements with low bandwidth.
As the automatic level of unmanned vehicles increases, memory technology shifts from the back seat of the car to the front seat; as memory technology is important for safety and performance. Historically, personal computers have been the engine of memory technology; but in the future, it is expect to become the main memory technology in the automotive industry. Today, some of the best automation platforms have already demonstrated this idea.
The Pegasus high-end computing platform, designed specifically for autonomous driving, recently announced that; it’s based on the latest performance and the latest DRAM technology. Overall, the Pegasus platform offers more than 1TB of memory per second to provide 5-level performance.
The importance of GDDR6 in future self driving cars
The bandwidth associated with GDDR6 provides a high level of automation with a track suitable for use in automobiles. Computing platforms with high memory bandwidth can continuously develop and improve independent learning algorithms. “What you’re seeing is a gradual improvement in the algorithm
“However, it is presented as a software update, such as getting regular updates for smartphone apps and operating systems.”
With the constant development of self driving cars, there are several versions with different features in the coming years. At this point, people need to have a clear understanding of when and to what extent automation exists and what responsibilities are associate with “implementation” and “monitoring” operations. The relationship between machinery and equipment must be carefully manage.
GDDR6 is a key technology that supports artificial intelligence computing engines and provides core memory throughput; that supports the ability of autonomous vehicles to drive responsibly and safely following NHSTA industry safety standards. GDDR6 is the best memory technology available today and has proven to work at very high temperatures and harsh conditions associated with automobiles.
Artificial intelligence is the basic technology needed to achieve autonomous driving. The ultimate computational performance required for autonomous artificial intelligence tools requires pioneering memory; And storage systems to process the large amounts of data that computers need to make human decisions. With more than 25 years of dedication to the automotive industry, where self driving cars need memory speeds.