What is occupancy grid mapping and how is it used in autonomous driving?

Occupancy grids represent the environment as a 2D or 3D grid where each cell contains the probability of being occupied. Construction: Sensor measurements update cell probabilities using Bayesian inference, with occupied cells having high probability and free space low probability. Sensor models translate range measurements to probability updates. Uses include: Path planning (identifying free space), obstacle avoidance, parking space detection, and sensor fusion (combining multiple sensor inputs in common representation). Advantages: Handles uncertainty naturally, doesn't require explicit object detection. Challenges include memory requirements for fine resolution over large areas and maintaining real-time updates. Dynamic occupancy grids track moving objects.