Various studies show that GPS data and a corresponding visualization system are sufficient to navigate a vehicle from A to B completely autonomously. Some maneuvers may have to be taught to the system in advance by driving the vehicle conventionally by a human over these passages, but not all routes need to be trained, because with the appropriate algorithm the vehicle can even learn from past maneuvers, as long as situations are similar enough - the system adapts. This will be increasingly expected from the automotive industry in the coming years. Many manufacturers are already relying on a variety of sensors such as radar, LiDAR, ultrasonic sensors, cameras, etc. and are fusing these into the vehicle's environmental perception.
In traffic, or rather in a constantly changing environment, there are many additional challenges to which humans still have the better trained instinct thanks to their perception of the environment (or to circumvent the liability issue) than a sophisticated system to react correctly, despite much higher reaction times. The SAE (Society of Automotive Engineers) has divided the driving modes into six levels, starting with autonomy level 0, where there is no automation, and ending with autonomy level 5: full automation. From autonomy level 3 upwards (conditional automation), the system takes control of the environment. Some manufacturers promise to offer autonomy level 5 soon, but common car models today are between autonomy level 2 and at most autonomy level 3.