Qualcomm is getting into the self-driving market, Ars Technica

        ******************************************(Qualcomm sign. Qualcomm******************** Qualcomm ************ the dominant provider of cellular modem chips, is the latest chip giant to seek a slice of the autonomous driving market. At the Consumer Electronics Show on Monday, Qualcommannounced its Snapdragon Ride platform, which uses chips derived from Qualcomm’s mobile products to help a vehicle drive.

It might seem strange for a smartphone chip company to tackle self-driving, but in reality, many of the same technologies exist in both products. Qualcomm has long leveraged its mobile chip lead to grab a significant chunk of the broader system-on-a-chip market. Smartphone makers buy Qualcomm Snapdragon chips that contain an ARM-based CPU, modem chip, and various support chips.

In recent years, Qualcomm and other mobile chipmakers have beenincludingstrongly powerful GPUs and dedicated AI chips in their SoC products — precisely the kind of silicon required for driving software based on machine learning. So it’s not much of a leap for Qualcomm to launch a new SoC that includes (as Qualcomm’s press release puts it) “high-performance multi-core CPUs, energy-efficient AI and computer vision (CV) engines, industry-leading GPU.

Insiders generally divide self-driving technology into two tiers. Some companies are working to build fully self-driving systems capable of operating without human intervention — though right now these are mostly prototypes being overseen by human safety drivers. Then there are advanced driver-assistance systems (ADAS). These are designed to drive in limited circumstances — most often freeways — and only under human supervision. Most major carmakers are already shipping ADAS systems, and they’re getting more sophisticated over time.

Qualcomm says the Snapdragon Ride platform includes more than just hardware. It will also ship with “safety-middleware, operating systems, and drivers” for the chips. In addition, Qualcomm will offer software for localization, perception, and behavior prediction — three key pieces in any autonomous driving system.

Qualcomm says it will focus on the ADAS market initially. The company says its new system can deliver 30 tera operations per second (TOPS) while drawing less power than rival chips. Qualcomm says it’s also planning to offer beefier systems that can deliver 130 TOPS that will be suitable for fully self-driving applications.

We expect low power consumption to be one of Qualcomm’s selling points. Power consumption is obviously a crucial characteristic of mobile devices, since customers expect them to work all day on a charge. Power budgets aren’t as tight on board a running car, but power consumption still matters. If Qualcomm can deliver adequate computing power while deploying some of the power-saving techniques developed in the mobile world, it could offer a compelling platform to carmakers.

However, Qualcomm’s competitors — most notably Nvidia and Intel-owned Mobileye — aren’t standing still. Nvidia’s expertise has long been in delivering massive amounts of computing power for its conventional graphics cards, making it well-positioned to serve the equally compute-hungry self-driving market.

Qualcomm’s announcement includes avague statementthat it will work with GM on ADAS systems, but so far there’s no mention of GM using the chips in any specific vehicles. Qualcomm says it’s aiming to have Snapdragon Ride chips in production cars in 01575879.                                                     

****************************************************** Read More) *************************************************