One of the main enablements of advanced driver-assistance systems (ADAS) and self-driving/autonomous cars is greater computing power. According to Intel, approximately 1 GB/s of data will need to be processed, stored and then analyzed quickly enough —in less than one second in a vehicle’s operating system — for a vehicle to react to changes in its surroundings.
Various ADAS technologies hold the promise of rectifying a variety of issues presently affecting drivers. As discussed in Bruce Franklin’s blog, autonomous driving requires systems to sense, perceive and act. And this requires processing power and memory. To understand what is driving data bandwidth and memory needs, let’s look at the requirements for a front camera system used in ADAS. (Note this in only one portion of an ADAS, there could be roughly six additional systems working in an autonomous ADAS.)
An ADAS front camera must capture and identify the object in front of a car, along with its distance, so the car can process and take the appropriate action. In a current standalone camera system, the resolution of the imaging sensor determines the furthest distance where objects can be detected and identified. In Mihir Mody’s article on ADAS front camera insights, a car with a 1 megapixel (Mp) resolution camera can detect a pedestrian at 34 meters. For a system based on an 8Mp resolution camera; a pedestrian can be detected from as far away as 101 meters.
Assuming an ADAS based on an 8Mp forward-looking camera with 16-bit depth that operates at a frame rate of 60 frames per second, data rate reaches almost 1 GB/s quickly. This data needs to be both stored and processed at line rate with maximum reliability across all temperature extremes.
Given the extreme temperatures cars operate in, both cold and hot, along with their reliability requirements, the memory solutions in this space need to meet a higher standard. Micron is helping to meet these requirements. For more information on how Micron is innovating and solving automotive memory requirements, visit our Automotive Memory Solutions page.