This week’s Client Electronics Present (CES) in Las Vegas has been much more dominated by automotive information than final 12 months, with scores of bulletins of latest in-vehicle growth platforms, automotive 5G companies, self-driving idea vehicles, automotive cockpit UIs, assisted driving techniques, and a bunch of electrical automobiles. We’ve additionally seen quite a few techniques that present Google Assistant or Alexa-driven in-vehicle interfaces akin to Anker’s Google Assistant based mostly Roav Bolt.
Right here we take a short have a look at a few of the main development-focused CES automotive bulletins up to now. The principally Linux-focused developments vary from Hyundai becoming a member of the Automotive Grade Linux mission to main self-driving or assisted ADAS platforms from Baidu, Intel, and Nvidia.
Hyundai jumps on AGL bandwagon
Simply previous to the launch of CES, the Linux Basis’s Automotive Grade Linux (AGL) mission introduced that South Korean automotive big Hyundai has joined the group as a Bronze member. The information follows final month’s addition of BearingPoint, BedRock Methods, Large Lake Software program, Cognomotiv, and Dellfer to the AGL mission. In October, AGL introduced seven different new members, together with its first Chinese language automobile producer — Sitech Electrical Automotive.
Hyundai’s membership doesn’t commit it to utilizing the group’s Unified Code Base (UCB) reference distribution for automotive in-vehicle infotainment, however it’s one other instance of the rising help for the open supply, Linux-based IVI stack. A number of main carmakers are members, together with Honda, Mazda, Mitsubishi Electrical, and Suzuki, but Toyota is the one AGL automotive producer to ship IVI techniques based mostly on UCB in most of its main fashions, from the Camry to its Lexus luxurious vehicles. In June, AGL introduced that Mercedes-Benz Vans was utilizing UCB for upcoming vans, and we will count on extra AGL commitments in 2019.
On the Westgate Resort Pavilion (sales space 1614) in Las Vegas this week, AGL is exhibiting off a 2019 Toyota RAV4 geared up with AGL techniques, and AGL members are providing demonstrations of AGL-based related automobile companies, audio improvements, instrument cluster, and safety options.
Baidu releases open supply Apollo three.5 self-driving software program
AGL is just not the one automotive mission providing an open supply answer. For the previous 12 months, Chinese language search and cloud big Baidu has been growing its Linux-driven Apollo stack for self-driving vehicles. At CES, it introduced Apollo three.5, with new help for “advanced city and suburban driving eventualities.” A hardware platform is offered with an Intel Core based mostly Neousys industrial pc geared up with an Nvidia graphics card, amongst different parts together with Baidu’s personal sensor fusion unit.
Baidu additionally introduced an Apollo Enterprise platform constructed on prime of Apollo designed for autonomous fleet operations. As well as, it revealed an open supply OpenEdge cloud-enabled edge computing platform with growth boards based mostly on NXP and Intel applied sciences. The latter is designed for in-car video analytics and incorporates Intel’s Mobileye know-how. Particulars have been sketchy, nonetheless.
At CES, Intel unveiled an Intel AV compute platform aimed toward autonomous vehicles. It contains a pair of Linux-driven Mobileye EyeQ5 sensor processing chips and a brand new Intel Atom 3xx4 CPU.
The Intel AV system gives 60 % better efficiency on the similar 30W consumption as Nvidia’s automotive centered Jetson Xavier processor, claims Intel, The Mobileye EyeQ5 processors are every claimed to generate 24 trillion deep studying operations per second (TOPS) at 10W every. Volkswagen and Nissan have introduced plans to make use of the sooner EyeQ4 processors when it launches later this 12 months. An EyeQ5 Linux SDK with help for OpenCL, deep studying deployment instruments, and adaptive AUTOSAR will probably be obtainable later this 12 months, and manufacturing will start in 2020.
The Atom 3xx4 chip, in the meantime, borrows high-end multi-threading and virtualization applied sciences from Intel’s Xeon processors for working totally different duties concurrently on totally different techniques across the automobile.
Nvidia Drive Autopilot
Intel is enjoying catchup with Nvidia within the autonomous automobile pc contest. Lately, Nvidia has more and more centered on the automotive enterprise, launching one of many first impartial self-driving automobile computer systems with its Drive PX Pegasus based mostly on its newly transport, octa-core Arm-based Jetson AGX Xavier module. At CES, it adopted up with a Xavier-based Nvidia Drive Autopilot system.
Not like the absolutely autonomous, “Stage 5” Drive PX Pegasus, the Drive Autopilot is designed for Stage 2 assisted ADAS techniques. As a result of ship in automobiles in 2020, the system contains a claimed 30 TOPS AI efficiency and gives “full encompass digicam sensor knowledge from outdoors the automobile and contained in the cabin.”
Drive Autopilot integrates a brand new Drive IX software program stack that may map and memorize typical routes to enhance efficiency sooner or later. It additionally gives driverless freeway merge, lane change, lane splits, and in addition to driver monitoring and AI copilot capabilities. We noticed no OS particulars, however presumably Drive Autopilot runs the Tegra4Linux stack used on different Xavier based mostly techniques.