Cleantech & EV'sNews

XPeng teases video of City Navigation Guided Pilot: Is it a better ADAS than Tesla FSD beta?

Ahead of its official release in China, XPeng Motors has posted a teaser video showcasing the engineering (beta) version of its City Navigation Guided Pilot (NGP). The one-take video shows XPeng’s P5 sedan driving autonomously from XPeng HQ in Guangzhou to the city center, demonstrating a number of ADAS functions. If and when it reaches mass production, City NGP could offer a legitimate challenge to Tesla FSD beta.

Chinese automaker XPeng Motors is starting to gain some momentum back in 2022, following a slower first half due to COVID-19 lockdowns and supply chain constraints. On July 1, the automaker reported 15,295 EVs delivered in June 2022, marking a 133% increase year over year.

Additionally, XPeng Motors delivered 68,983 EVs in the first half of this year, a YOY increase of 124%, which, according to the automaker, was enough to secure its crown as number one in deliveries “among emerging auto brands in China” for four consecutive quarters.

We got the chance to test drive XPeng’s P5 and P7 sedans last month in the Netherlands and experienced some of the automaker’s ADAS functions first hand. While many of those features are still being adapted for EU roads and regulations, features like XPeng’s Navigation Guided Pilot (NGP) are already in operation in China.

For example, the Highway version of XPeng’s NGP was launched in China in Q1 of 2021 on the premium version of the P7, which is supported by XPILOT 3.0 software. This ADAS feature functions without LiDAR, but its potential is limited – it allows for assisted driving on highways and expressways but not for city-level driving.

XPeng first teased its City NGP in a video last September, showing a beta version of the software navigate busy urban streets without human interference. As City NGP gets closer to mass implementation, XPeng has shared a new video further demonstrating that this ADAS could be a genuine competitor to Tesla’s polarizing FSD beta.

XPeng ADAS
Map of the XPeng P5 route while it tested City NGP ADAS / Source: XPeng Motors

XPeng video demonstrates impressive ADAS maneuvers.

The teaser video, which was recently tweeted by XPeng Motors, features over four minutes of uncut footage as a P5 sedan equipped with City NGP travels 26km (~16 mi), performing a multitude of ADAS functions autonomously along the way.

This includes cruising, following, navigating and changing lanes, overtaking and changing lanes, entering and exiting roads, and detouring stationary vehicles and objects – all while maintaining an appropriate speed. 

In the video below, the P5 EV recognizes traffic lights and stops, automatically changes lanes, turns left and right at intersections, moves through roundabouts and viaducts, and passes through five tunnels. All the while, XPeng’s ADAS safely avoids road construction, pedestrians, and cyclists.

These abilities are thanks in part to XPeng’s XPILOT 3.5 software, which will debut on the P5 sedan alongside another production first for the automaker – LiDAR. These two sensors are vital in delivering perception fusion to the P5 which integrates camera, radar, and LiDAR together instead of relying on one single vision perception, like Tesla’s FSD, for instance. The following hardware will be required to support City NGP when it is approved for roads in China:

  • Two LiDAR units
  • Five millimeter wave radars
  • Thirteen vision cameras
  • IMU-GPS

XPeng’s autonomous driving approach looks promising.

According to XPeng Motors, its XPILOT is the first full-stack self-developed software to be mass produced in China and the strongest ADAS architecture to boot. The multi-sensor fusion positioning based around camera vision, high-precision maps, GPS, IMU, and wheel speedometers allows for city positioning accuracy at the centimeter-level without having to rely on point cloud maps.

As a multi-perception fusion solution, City NGP can detect, classify, and locate more types of objects with an accuracy up to the decimeter level and more accurately judge the position and speed of objects while understanding the orientation and posture of pedestrians.

If and when XPeng successfully brings City NGP into mass production, its full stack of self-developed ADAS will support the full spectrum of driving scenarios (high speed highways, city level medium speeds, and ultra-low speed parking lot scenarios). Furthermore, XPeng believes it will benchmark Level 4 ADAS for mass produced EV models as the first to deliver urban navigation assistance in China.

According to XPeng, it plans to release its new City Navigation Guided Pilot after obtaining relevant approvals in China. Here’s the teaser video for you to see it in action.

Electrek’s Take

Although XPeng Motors calls this its “engineering version” of City NGP, it’s just a different way of saying beta ADAS because that’s exactly what it is. Although it has not received full regulatory approval in China to operate in the P5, it looks damn promising as a genuine threat to Tesla FSD beta.

Time and again we’ve listened to Elon Musk punt the FSD delivery timeline down the field, turning the most loyal of fans into sad clowns that truly believe the technology is still “just weeks away.” I stopped holding my breath years ago but am honestly still rooting for the technology.

Such delays have led people like myself to set our sights elsewhere to see who else can scale this technology. XPeng appears up to the task, taking a kitchen sink approach with an arsenal of sensors that combine to analyze hectic urban road environments in real time.

Meanwhile, Tesla continues to work the kinks out on FSD beta. I truly believe we’re living in a world where both approaches to autonomous driving can work in mass produced EVs, but I personally feel like XPeng’s NGP technology is the Hansel to Tesla FSD’s Zoolander. What do you think? Be nice.


Subscribe to Electrek on YouTube for exclusive videos and subscribe to the podcast.


Author: Scooter Doll
Source: Electrek

Related posts
AI & RoboticsNews

Nvidia and DataStax just made generative AI smarter and leaner — here’s how

AI & RoboticsNews

OpenAI opens up its most powerful model, o1, to third-party developers

AI & RoboticsNews

UAE’s Falcon 3 challenges open-source leaders amid surging demand for small AI models

DefenseNews

Army, Navy conduct key hypersonic missile test

Sign up for our Newsletter and
stay informed!