Perception & Mapping
2D SLAM Visualizer
A live 2D SLAM simulator: a differential-drive robot explores one of six environments while its LiDAR, wheel odometry, occupancy-grid mapping, correlative scan matching, and pose-graph loop-closure optimization all run in the open. Ground truth, odometry, and the SLAM estimate are tracked as three separate poses so you can watch odometry drift and see how loop closure snaps a distorted trajectory and map back into a globally consistent one.
Loading SLAM visualizer…
Mathematical Formulation & Engineering Notes
Occupancy update: Lₜ(mᵢ) = Lₜ₋₁(mᵢ) + inverse-sensor-model (log-odds). Loop closure adds a pose-graph constraint, resolved via iterative Gauss-Seidel relaxation over all odometry + loop-closure edges.