{"id":34478,"date":"2022-10-24T16:26:15","date_gmt":"2022-10-24T20:26:15","guid":{"rendered":"https:\/\/autosector.com\/?p=34478"},"modified":"2022-10-24T16:26:15","modified_gmt":"2022-10-24T20:26:15","slug":"xpeng-1024-tech-day-2022-recap-autonomous-driving-robotaxis-and-flying-car","status":"publish","type":"post","link":"http:\/\/autosector.com\/?p=34478","title":{"rendered":"XPeng 1024 Tech Day 2022 recap: Neural net autonomous driving, robotaxis, and ‘flying car’ footage"},"content":{"rendered":"

XPeng completed its fourth annual 1024 Tech Day in China today, highlighting several of its latest technological breakthroughs in electric mobility. The presentation included several advancements in autonomous driving, robotic animals, and self-evolving AI platforms. Additionally, XPeng Inc.<\/a> shared that its sixth generation eVTOL \u2013 the closest thing we\u2019ve seen to an actual flying car \u2013 completed its maiden flight. Lots to unfold here, so let\u2019s get started.<\/p>\n

\n

Table of contents<\/h2>\n<\/div>\n

Quick recap leading up to XPeng\u2019s Tech Day 2022<\/h2>\n

XPeng Inc. ($XPEV)<\/a> is a Chinese EV manufacturer founded in 2014 in Guangzhou that is made up of XPeng Motors<\/a> and XPeng AeroHT, formerly known as XPeng Huitian \u2013 which specializes in Urban Air Mobility (UAM).<\/p>\n

In addition to manufacturing EVs<\/a> and electric vertical takeoff and landing (eVTOL) vehicles<\/a>, XPeng develops its own software and charging products and has begun dipping into other ventures like robotics as well.<\/p>\n

During last year\u2019s 1024 Tech Day, XPeng revealed advancements to its XPILOT ADAS<\/a>, \u201cX Power\u201d superchargers, and a robotic pony prototype<\/a> equipped with LiDAR. XPeng wrapped last year\u2019s Tech Day with what was probably its coolest announcement: the debut of its sixth generation eVTOL \u2013 the first to actually be worthy of the term \u201cflying car.\u201d<\/p>\n

Previous versions of its eVTOLs<\/a> didn\u2019t have wheels but were based around the XPeng\u2019s P7 EV architecture, so the UAM division still used the term \u201cflying car.\u201d However, AeroHT\u2019s latest model can traverse both roads and air and has recently completed its maiden flight. More on that below.<\/p>\n

\"XPeng<\/figure>\n

Big upgrades to XPeng\u2019s XNGP ADAS<\/h2>\n

When it debuted on the automaker\u2019s G9 SUV<\/a> in September, XNGP (navigation-guided pilot) was announced as XPeng\u2019s last step before achieving fully autonomous driving. XNGP combines all scenarios of the automaker\u2019s existing ADAS capabilities<\/a> (highway, city, parking) into one holistic system that will soon no longer require high-precision maps to function \u2013 essentially opening up its availability to any and all areas.<\/p>\n

At Tech Day 2022, XPeng shared that XNGP is backed by major hardware upgrades, including 508 TOPS of computing power, a dual-LiDAR system, 8-megapixel HD cameras, and a new software architecture called XNet, which operates using a closed-loop, self-evolving AI and data system.<\/p>\n

XPeng unveils XNet neural network-based architecture at Tech Day<\/h3>\n

XPeng\u2019s XNet varies from its first-generation visual perception architecture by adopting a deep neural network that was developed in-house to deliver visual recognition with human-like decision-making capabilities, drawing data from multiple cameras. <\/p>\n

The company explained that its autonomous driving technical stack can reach 600 PFLOPS, increasing the training efficiency of the autonomous driving model by over 600 times. To that note, model training can be significantly reduced from 276 days to just 11 hours.<\/p>\n

For added texture in regard to XNet\u2019s streamlined efficiencies, it now only uses 9% of its Orin-X chip\u2019s processing power, compared with 122% before optimization. These upgrades have enabled XPeng to establish an entirely closed-loop autonomous driving data system (data collection, labeling, training, and deployment) that utilizes lightning-fast machine learning to consistently self-improve. Per the release:<\/p>\n

\n

XPeng\u2019s high-efficient AI capabilities enable consistent and unsupervised machine leaning and rapid iterations in training models, resolving over 1,000 rare corner cases each year. This highly efficient closed-loop AI and data system has helped reducing incident rate for the Highway NGP by 95%.<\/p>\n<\/blockquote>\n

Continuing through the Tech Day presentation, the XPeng team explained that the technology above will help enable the automaker to develop and deploy an autonomous robotaxi fleet in China.<\/p>\n

\"XPeng<\/figure>\n

G9 SUV will be part of the XPeng\u2019s robotaxi network in China<\/h2>\n

On stage at Tech Day 2022 in Guangzhou was XPeng\u2019s vice president of autonomous driving, Dr. Xinzhou Wu, who shared that XPeng is developing a robotaxi network<\/a> in China. The company shared that the G9 SUV became the country\u2019s first mass-produced commercial vehicle to pass the Autonomous Driving Closed-field Test led by the Chinese government.<\/p>\n

By completing the test to operate in a tier-one city like Guangzhou, XPeng states it has taken a major step forward in robotaxi development, all while using existing technology present on its latest EV. Wu elaborated:<\/p>\n

\n

Obtaining the road test permit by our mass-produced commercial vehicles \u2013 with no retrofit \u2013 is a major achievement. Our platform-based robotaxi development aims to generate significant cost benefits, and ensure product quality, safety and user experience.<\/p>\n<\/blockquote>\n

\"\"<\/figure>\n

XPeng\u2019s robotic pet gets an upgrade to support better motion<\/h2>\n

As you can see from the image above, XPeng\u2019s software is not the only technology that got upgrades since last year\u2019s 1024 Tech Day event. The aforementioned \u201cpony\u201d robot has gotten a revamped design to support multidegree-of-freedom (MDOF) motion and locomotion capabilities. Other upgraded features include the following:<\/p>\n