P7/G7 Car Full-scene VLA Function to Go Online Soon: "Ghost Head" for Blind Spot Detection and Predictive Braking

Phoenix Technology News, August 22nd. Xiaopeng announced today that the full-scene VLA function will be launched on Xiaopeng P7 and G7 two car models.

According to introduction, Xiaopeng's full-scene VLA function can recognize yield signs and control speed, maintain a safe distance during stops, predict and defend against blind spots, and identify potential risks when encountering incidents. The vehicle will slow down in advance, control speed at road intersections, identify extreme weather conditions, recognize flooded roads, and prioritize efficient travel during non-restricted hours.

As reported, in the field of autonomous driving, the VLA model can extract environmental information from sensor data and understand human instructions through language models to generate a clear decision-making process. It converts multi-modal information into specific driving instructions.

As a reference, Xiaopeng P7 is equipped with an 800V power architecture, 5C batteries, and can accelerate from 0-100km/h in 3.7 seconds. The car features three Turing AI chips and has an effective computing power of 2250 Tops. The price has not been announced, and the new car will be available on August 27th.