Automakers have different approaches to how many and what types of sensors should be used for autonomous driving, with the most advanced vehicles using 30 sensors. OEMs are still figuring out which sensors are best for different purposes, with lidar providing excellent distance information but not working well in bad weather and cameras providing poor distance information and requiring substantial computing resources. AI deep learning techniques are needed for pattern recognition and route planning to address these limitations and optimise driving decision making. Sensor fusion, which combines multiple types of sensors, is also important to achieve safe autonomous driving. The use of augmenting technologies such as V2X, 5G, advanced digital mapping and GPS in smart infrastructure may enable autonomous driving with fewer onboard sensors.
https://semiengineering.com/how-many-sensors-for-autonomous-driving/