Tesla FSD vs China’s Xpeng XNGP Autonomous Driving Test

What is the best autonomous driving system on the market? Tesla thinks the answer is its FSD Beta (Full Self Driving). However, Chinese electric vehicle startup Xpeng proposed its XNGP (Navigation Guided Pilot), so we’re here to show you a comparison between the two.

Both Tesla’s FSD Beta and Xpeng’s XNGP offer Level 2 semi-autonomous driving, which means they still require human driver monitoring—Level 5 would represent fully autonomous driving.

Other carmakers offer such Level 2 ADAS (Advanced Driver Assistance Systems). Nevertheless, with Tesla being the most vocal and XPENG having set its sights on the US company, we’ll start by showcasing the differences between the two.

Camera-only vs. cameras+sensors

After multiple changes, Tesla’s latest FSD (hardware 3.0 and software) relies solely on its 9 cameras, foregoing any sensors, as far as perception goes.

XNGP, on the other hand, uses 31 pieces of hardware for perception. Vision is still the dominant feature, which is why the system has 11 cameras. These are aided by 2 LIDAR units, 12 ultrasonic radars, 5 mm wave radars and a high precision positioning unit.

For the record, cameras, which provide high-definition imagery, are the only ones that can rear road markings and traffic signs. In the dark, LIDAR is superior to cameras and while radar is less precise than LIDAR, this is unaffected by weather.

While FSD doesn’t use HD maps—showcasing details like curbs and traffic signs—Xpeng is now offering a mapless version of XNGP in more and more cities.

Another important difference comes in terms of the computing power. Sure, both FSD and XNGP rely on centralized AI software running on supercomputers (Tesla Dojo and Xnet 2.0), which helps with redundancy, among others.

But while Tesla drivers only need to pay for a software update to install FSD Beta (on a standard chip with 144 TOPS of computing power), you need to order XNGP with the optional LIDAR sensors and an Nvidia Drive Orin chip (508 TOPS).

Here’s BYD’s Yangwang U8 electric SUV mastering parallel parking by sliding its posterior in

Tesla FSD is better on the highway, Xpeng’s XNGP has the edge in the city

It’s important to note that, at the time of writing, Tesla’s FSD is available in the US, but not China or Europe (the Old Continent should get FSD later in 2024). Founded in 2014, Xpeng doesn’t sell vehicles in the US but is active in certain European countries and Israel (as of 2023). However, XNGP has yet to be offered to customers outside China.

In the said context, this FSD vs XNGP test is as close to a direct comparison as possible. The test comes from vlogger Ethan of the Wheelsboy YouTube channel, an American who currently spends most of his time in China reviewing local cars.

In November 2023, the vlogger flew from China to California, where he drove a Tesla Model 3 with FSD Beta. This was a 2018 model, but since it packed the latest hardware and software mentioned above, it essentially offered the same Level 2 ADAS as a 2023 model. The drive included highways, byways, as well as city traffic (mostly in San Francisco).

Also last year, the man drove a 2024 Xpeng G9 on Chinese highways and byways, as well as in Guangzhou—Xpeng’s home city—and Shanghai.

On the highway, both Tesla FSD and Xpeng’s XNGP maintain a safe distance from the car in front by using smooth brake inputs and can perform automatic overtaking. However, FSD seems more natural than XNGP, especially while navigating transitional roads between highways.

In the city, XNGP feels more confident than FSD, especially in complex situations, as is the case with making left turns and being around pedestrians and cyclists. However, the advantage of the Chinese ADAS is mostly owed to its reliance on HD maps (as tested), so it remains to be seen how the expanding mapless version will perform.

Interestingly, despite XNGP appearing to have an edge over FSD in urban driving, it still made decisions that caused the driver to feel unsafe—in the heavy-duty Shanghai traffic—at least once.

Nevertheless, FSD’s Autosteer feels more precise in cities, thanks to requiring a bit more effort while keeping your hands on the wheel. In contrast, the lesser effort that comes with XNGP means you can unintentionally cancel a lane change and therefore exit the ADAS for a few seconds.

It’s also worth mentioning that while San Franciso traffic is heavy by US standards, traffic in Shanghai is considerably more complex, not least thanks to the presence of more cyclists. These (semi-)autonomous driving systems may also change the way we shape our roads. For one, the driver noticed that XNGP was inferior to FSD on the highway when it came to reading road signs (it kept slowing for lower speed limits at exits). But this could be due to road signs in China being all in the same plane and therefore potentially confusing the ADAS.

Here’s how many people chose the Range Extender on the Tesla Cybertruck

How long before you’re allowed to use Tesla FSD after a ban?

The final chapter of the FSD vs XNGP comparo involves driver monitoring. Tesla claims it uses torque sensors in the steering wheel and, allegedly, the in-car camera. However, there have been countless cases of people putting teddy bears in the driver’s seat without the ADAS disengaging.

XNGP mixes steering wheel torque sensors with an eyeline camera and, during this test, it proved effective.

Now, repeated offenses will see XNGP downgrading the driver (i.e., you have to watch an instructional video and use the system in limited “rookie” mode). With FSD, the penalty for not paying attention on multiple occasions involves getting a temporary ban from using the system—being allowed back in can take anywhere from a few weeks to a few months.

Even with all their structural differences, Tesla FSD and Xpeng’s XNGP offer remarkably similar features. And we’re looking forward to seeing other brands joining the fight at the customer-available level—for one, Mercedes’ Drive Pilot aims to be approved as the first Level 3 autonomous driving system available to US buyers.



Please enter your comment!
Please enter your name here