Xpeng Motors Announces its Next-Gen 5G-Powered Autonomous Driving Architecture

author: Eric Walz   

He Xiaopeng, Chairman & CEO of Xpeng Motors, speaks at Auto Guangzhou 2020.

Electric vehicle startup Xpeng Motors is emerging as a strong competitor to Tesla in China with its advanced battery-powered vehicles. In addition to building advanced EVs, the company is focused on autonomous driving and vehicle connectivity.

The latest result of that work was announced at Auto Guangzhou 2020, when Xpeng announced its next-generation autonomous driving architecture. Xpeng calls its autonomous driving system XPILOT.

XPILOT is offered in the company's new P7 Smart Sedan, which is a strong competitor to the Tesla Model S in China. Deliveries of the P7 began in June. The latest version of XPILOT will bring several autonomous driving features to Chinese customers for the first time, including Level 3 autonomous driving functions for highways, urban roads and automated parking. 

In a presentation at Auto Guangzhou 2020, Xpeng announced that it made significant improvements to its autonomous driving software and hardware systems for its 2021 models, including integrating lidar in its vehicles during production to support automated driving. 

Xpeng is the first automaker to add lidar to its vehicle during production. The lidar will significantly improve the vehicle's high-precision object recognition performance, according to Xpeng.

Lidar works by bouncing pulses of laser light off objects and measuring the time its takes for the beams to reflect back. The reflected light is used to generate a precise, three-dimensional image of a vehicle's surroundings known as a "point cloud."

Over the past several years, lidar has become an essential piece of hardware for self-driving cars, acting as the "eyes" of the vehicle. Lidar perceptions systems are combined with other hardware and sensors to help a self-driving vehicle to safely navigate. 

"Introducing Lidar technology into production vehicles is a breakthrough in popularizing autonomous driving, and an endorsement of our in-house R&D process. Our customers will benefit from this premium advanced technology, which makes autonomous driving more driver-friendly, safe and effective," said He Xiaopeng, Chairman & CEO of Xpeng.

In addition to lidar, Xpeng's new architecture for autonomous driving includes high-definition (HD) cameras, millimeter-wave radars, ultrasonic sensors and high-precision GPS and mapping systems. The hardware is all powered by a high-performance computing platform and runs on Nvidia's DRIVE Xavier SoC.

The P7 is the first production vehicle in China powered by Nvidia's powerful Xavier SoC.

Nvidia's DRIVE AGX platform is a scable, open platform that serves as the AI-powered "brain" of a self-driving vehicle. Xavier is the first production-level SoC dedicated to supporting highly automated and autonomous driving functions. Xpeng is using NVIDIA Xavier's Deep Learning Accelerators for AI inference, which is used to support Level 2 and higher autonomous driving. 

Nvidia's Xavier SoC is also a powerhouse delivering 30 TOPS (trillions of operations per second) of performance while consuming just 30 watts of power. 

Xpeng's P7 Smart Sedan comes packed with hardware for automated driving with a total of 32 sensors. The entire hardware suit includes redundancy to avoid any single point failure. It will also cover more usage scenarios, effectively enabling Xpeng vehicles to perform low-speed Navigation Guided Pilot (NGP) functions for city driving.

Xpeng developed an integrated control unit the company calls "XPU" to support its autonomous driving hardware. The XPU control unit integrates vehicle control and automated parking functions, making decision-making and execution more efficient. At the same time, its streamlines the number of controllers and wiring harnesses required, so that the overall weight of the system is reduced.

Xpeng said the software architecture is based on a full-stack of in-house-developed technologies, consisting of multiple modules for perception, behavior planning, behavior/motion prediction and map fusion. 

The autonomous driving technology will use 5G connections speeds. Xpeng employs a real-time kinematic (RTK) terminal that offers a higher degree of accuracy for maps and GPS to support autonomous driving in both rural and urban areas, the company said.

The adoption of Lidar technology provides a perception capability that is unaffected by lighting conditions, meaning it performs equally well in bright sunlight or complete darkness.

By combining the lidar with millimeter wave radar ultrasonic sensors Xpeng says it delivers "near-perfect perception performance."

Xpeng's XPILOT is also being designed with the capacity to further increase computing power, supporting the implementation of even higher-level autonomous driving functions. In the future, Xpeng said the high-performance computing platform will also be upgraded from 30 TOPS to over 200 TOPS, representing nearly tenfold increase in computing power to several hundred trillion operations per second. 

The next-generation autonomous driving architecture will achieve an advanced level of autonomous driving, with wide coverage of different road conditions and longer continuous autonomous driving times, the company claims.

XPeng's XPilot was specifically developed for China's unique road conditions, with automated features adapted for China's crowded urban roadways, including a unique "low speed follow function", which reduces the chances of aggressive drivers cutting in front of the vehicle in traffic.

The next-generation system also covers multiple weather conditions and supports city and highway driving, expressways, underground parking lots, toll gates, tunnels and night driving conditions.

Xpeng Motors prides itself on developing all of its advanced autonomous driving technology in-house. More than half of the company engineering teams work in R&D to support the development of autonomous driving. 

The company also developed an AI-powered intelligent OS for its vehicles called XSmart OS.

Eric Walz

Originally hailing from New Jersey, Eric is a automotive & technology reporter covering the high-tech industry here in Silicon Valley. He has over 15 years of automotive experience and a bachelors degree in computer science. These skills, combined with technical writing and news reporting, allows him to fully understand and identify new and innovative technologies in the auto industry and beyond. He has worked at Uber on self-driving cars and as a technical writer, helping people to understand and work with technology.