All Autonomous Vehicles use advanced AI to help them recognize, people, cars, objects, etc, as well as helping them make decisions on the best course of action. AI is trained by giving thousands of examples of images and videos in order for the AI to be able to understand the road and the driving environment.
The amount of data that needs to be collected for training material is vast and time consuming to collect. Companies like Google, try to collect the data themselves by sending their own fleet of vehicles equipped with cameras and other sensors to drive the planet. This is a very costly process.
Unlike humans who can use a normal map to navigate, autonomous vehicles need a very accurate HD map in order to accurately position themselves and to help understand the road network around them.
The typical map data that people currently use in their navigation apps is accurate to around 3m, whereas HD maps for Autonomous vehicles are accurate to less than 20cms. Mapping the world at that resolution takes a lot of effort and requires a huge amount of data.