ADAS and other autonomous driving technologies can apply automatic braking, park your car, adapt cruise control to road conditions, monitor blind spots, watch out for pedestrians, and more. These advanced functions are the result of innovations in embedded technology and chip design.
To achieve the very best in this technology, however, chips must be designed specifically for self-driving cars or advanced driver-assist systems, and there are plenty of chipmakers doing just that: it’s a market .
South Korean media has recently reported a partnership between Samsung and Google (Alphabet) in which .
Waymo is Google's autonomous ride-hailing service. Image used courtesy of
The will see the South Korean tech giant design and supply ICs that control all functions of the vehicle while computing data collected by its many sensors. The car will also communicate and exchange data with the Google data center in real-time.
While details are scant at this time, Samsung is expected to leverage the very same state-of-the-art design and manufacturing processes that it has used to build chips on the 7nm and 5nm process nodes.
Qualcomm, the chipmaker that produces chips used in millions of smartphones, announced at the start of 2020 that it optimistically claimed could be deployed in cars as soon as 2023.
Known as Snapdragon Ride, the platform solution is designed specifically for autonomous driving and ADAS and is supported by three core “pillars:” Snapdragon Ride Safety SoCs, a Snapdragon Ride Safety Accelerator, and a Snapdragon Ride Autonomous Stack.
Built on a 5nm process, the platform is said to offer sensor fusion and road world visualization. Screenshot used courtesy of
The platform is built upon modular multi-core CPUs, artificial intelligence and computer vision engines, and GPUs. This enables it to support all , from L1/L2 (driver assistance/partial automation, which we currently have in some vehicles) all the way to fully autonomous driving at L4/L5 (high/full automation).
NXP Semiconductor is one of the largest providers of automotive chips and currently accounts for around 11% of the total market. While a large portion of these chips are used in applications like entertainment, infotainment, and in-vehicle networking systems, some are used in applications critical for self-driving such as radar, battery management, and advanced driver-assist systems (ADAS).
NXP’s radar technology provides a scalable portfolio of integrated and secure products that are already used in real-world applications to aid with tasks like lane changing, parking, autonomous emergency braking, and adaptive cruise control. It uses several different chips to handle everything from power management (, ) to high-performance radar imaging ().
Block diagram of the PF5200 dual-channel PMIC. Image used courtesy of
NXP’s strength in the market is perhaps best illustrated by Qualcomm’s 2018 bid to buy the Dutch chipmaker. Regulatory issues derailed these plans, however.
GlobalFoundries recently announced a .
As part of the deal, GlobalFoundries will develop high-frequency radar chips for Bosch at its Fab 1 facility in Dresden, Germany. The chips will be designed to operate at a much higher frequency than previous generations to help the radar detect objects that are farther away at a higher degree of accuracy than lower-frequency radar chips seen in current vehicles.
Cleanroom of GlobalFoundries' Fab 1 in Dresden, Germany. Image used courtesy of
GlobalFoundries said the chips are being targeted for delivery in the latter half of this year.
Chips as the Brains of ADAS
Chips act as the “brains” of advanced driver-assist systems, without which current autonomous driving achievements and tomorrow’s “self-driving” cars could not be feasible. This is exactly why the current ; they’re an essential component in virtually every modern car, even at the most basic level.
However, many key chipmakers—including Inter, SK Hynix, and GlobalFoundries—have announced plans to .
The article comes from the Internet. If there is any infringement, please contact [email protected] to delete it.