Although driving may seem simple to humans, it remains one of the most challenging tasks for machines, as countless new scenarios can arise while on the road.

Now, we can already be in San FranciscoOrWuhanSeeing self-driving cars on the road in various locations. Under GoogleCurrently, Waymo provides over 200,000 paid self-driving taxi rides per week through its operations centers in Los Angeles, San Francisco, and Phoenix.2025 is also expected to be a pivotal year for the development of self-driving trucks, as several companies plan to launch commercial operations in the U.S.

Despite initial progress, autonomous driving still faces numerous challenges, including regulatory complexities, user acceptance issues, and technological barriers. Artificial intelligence (AI)—particularly the emerging capabilities of generative AI—are emerging as key technologies that could help address some of these hurdles.

The Role of Generative AI in the Driving Process

So, how do vehicles "think"? And what role does generative AI play in this process? The following three aspects will help answer these questions:

1. Artificial Intelligence"Reimagining" the car's "brain" with an end-to-end model

Traditional autonomous driving systems typically rely on rule-based decision-making. While these systems offer predictability and transparency, they clearly fall short when it comes to handling the complex, real-world driving conditions.

A comparison of rule-based models and artificial intelligence models.‍‍‍‍

Image source:World Economic Forum

Artificial intelligence is helping to overcome these shortcomings,Initially introduced into various subsystems, it has now evolved intoEnd-to-End (E2E) Artificial Intelligence ModelThese end-to-end AI models integrate perception, prediction, and planning into a single neural network,Enabling vehicles to learn and make decisions with unprecedented speed and agility.

However, E2E models also present several challenges, the most prominent of which is the issue of interpretability—since the models exhibit "black-box" characteristics, this raises significant safety concerns. Fortunately, recent breakthroughs are paving the way for more interpretable and verifiable solutions.

Autonomous Driving Technology CompanyWaabiRaquel Urtasun：“AI，，，”

“，”

2. ：“”

AI，，，

AI，

，WaymoWaabi SimulyticAI，

‍‍‍‍

Image source:，

SimulyticAndrea Kollmorgen“，，”

，，

3.

，，，，（DMS）（HMI），AI

DMS，AIHMI，AI，，

Erhan Köseoğlu：“‘ABC’——AI（Big Data）（computing power），DMSHMI，，AI”

AI

AI，，，

，AI，

，AI，AI，

· (Alwin Bakkenes)：“，AI，，‘’”

“，UI/UXAI，，”

，，

TIERIVAutoware Shinpei Kato：“，”

，，AI，AI

Agenda ，，