It is reported by XPeng, recently, in accordance with the relevant requirements of the autonomous driving function test of intelligent networked vehicles in Guangzhou, XPeng G9 successfully passed the closed field test of autonomous driving, becoming the first mass-produced vehicle in China to pass the test of closed field of autonomous driving.

The test was performed at night. The autonomous driving performance of vehicles in complex comprehensive urban road scenes was verified through 31 urban road scene tests including identification and response of various indication signals, conflicting driving of vehicles turning left and right, overtaking in parallel, two-wheeled vehicles and pedestrian avoidance.

*Avoid crossing pedestrians: In urban road scenes, pedestrian crossing is a common emergency situation. In this scene, multi-sensor fusion can be used to quickly identify pedestrians, give courtesy, and then move on.

In all test scenarios, XPeng G9 completed the driving operation safely and accurately within the scope of traffic regulations. It has become the first mass-produced model in the industry to successfully pass the test by upgrading and strengthening the autonomous driving software without changing the hardware.

*Avoidance of two-wheeled vehicles: takeaways and couriers make urban life more and more convenient, and also bring more challenges to intelligent assisted driving. In this scenario, the suddenness and the speed of incoming vehicles require higher technology coping ability.

The biggest difference from other models on the market that has passed this test is that the XPeng G9 test vehicle does not have any additional sensors and computing units, and uses sensing components such as cameras, millimeter-wave radars, and lidars that are entirely based on the mass-produced vehicle itself. The 360° perception system, which realizes the hardware level of autonomous driving is exactly the same as that of mass-produced models.

* Emergency braking of the car in front: Congested urban traffic often encounters an emergency situation ahead, which may be a misoperation by the driver of the car in front, or a “ghost probe” suddenly found in front of the car. The G9 test vehicle can calculate, plan and process based on the speed and distance of the vehicle in front.

As XPeng’s fourth mass-produced model, the Xpeng G9 launched in September is positioned as an “ultra-fast charging full-smart SUV”, with a fully upgraded hardware configuration and equipped with Xpeng’s second-generation assisted driving system XNGP. It will realize the ultimate product form of assisted driving — full-scene assisted driving.

* Cone-bucket detour: In scenes such as urban road maintenance, greening workers’ watering operations, etc., the cone-bucket usually appears to warn. The G9 test vehicle can automatically detect cones by integrating multi-sensor data, and plan detours and avoidance.

In the field of autonomous driving, XPeng has reached an industry-leading level. Over the past few years, based on the first-generation assisted driving system XPILOT, XPeng has successively implemented high-level assisted driving functions including APA intelligent parking, high-speed NGP intelligent navigation assisted driving, and VPA parking lot memory parking. Further it also launched the urban NGP function in mass production, taking the lead in the second half of assisted driving.

With the delivery of the G9, its second-generation assisted driving system, XNGP, will begin to realize full-scene intelligent assisted driving capabilities in most cities in 2023, and gradually complete the driving and parking functions in 2024, forming a complete point-to-point intelligent navigation assisted driving.

* Vehicle conflict: When driving at an unprotected intersection, there may be vehicles passing in all directions. The G9 test vehicle automatically judges the relationship between the position and speed of the vehicle and other vehicles, and makes optimal processing.

After the passing of this test, XPeng will continue to gradually explore in the field of intelligent travel business, and complete the accumulation of more driving data and scenarios, enabling high-level intelligent assisted driving and even mass production of unmanned driving through data closed-loop.

* Note: The road scenes presented here are all shot on closed roads for professional testing.