Occupancy Grid Mapping: Theory and Log-Odds Update Rules for Robot Perception

Occupancy grid mapping represents an environment as a lattice of cells, each storing the probability that the corresponding space is occupied, which lets robots handle noisy range sensors without forcing early binary decisions. Each sensor measurement is transformed into the map frame using the robot's pose and sensor calibration, and the ray from the sensor to the detected surface marks traversed cells as likely free and the endpoint as likely occupied. To make repeated probabilistic updates efficient, implementations commonly use log-odds, where independent evidence is accumulated by simple addition. The inverse sensor model governs how each measurement changes cells, and ray traversal methods identify which cells each beam crosses, a computationally heavy step for LiDAR scans. Practical systems also bound occupancy probabilities so that repeated observations do not drive values toward absolute certainty.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)
Log in to join the discussion and vote.
Log in