How ivanwong9290/cmu_16833_slam works
This repository is a collection of four SLAM coursework implementations, not one integrated application. Its principal workflows estimate robot pose and maps from motion and sensor data: particle-filter localization, EKF SLAM, linear/nonlinear graph-based SLAM, and dense RGB-D mapping with ICP and point-based fusion. The graph preserves these as separate product branches. Source evidence is uneven: EKF and some solver or particle-filter internals were not excerpted, so uncertain connections are omitted or marked unverified rather than inferred as certain.
Main entry points: icp_fusion/main.py RGB-D mapping runner · code/main.py Localization runner · preprocess.py Pose and normal preprocessing · transforms.py Camera transforms
Components 16
Particle Filter
- Localization runner
particle_filter/code/main.py - Occupancy map
particle_filter/code/map_reader.py - Motion model
particle_filter/code/motion_model.py - Laser sensor model
particle_filter/code/sensor_model.py - Particle resampling
particle_filter/code/resampling.py
EKF SLAM
- EKF SLAM
ekf_slam/code/ekf_slam.py
SLAM Solvers
- Linear SLAM
slam_solvers/code/linear.py - Nonlinear SLAM
slam_solvers/code/nonlinear.py - Sparse solver methods
slam_solvers/code/solvers.py - State and plotting utilities
slam_solvers/code/utils.py
Dense Mapping
- RGB-D mapping runner
icp_fusion/main.py - Pose and normal preprocessing
icp_fusion/preprocess.py - Camera transforms
icp_fusion/transforms.py - ICP alignment
icp_fusion/icp.py - Point-based fusion
icp_fusion/fusion.py - Point-cloud utilities
icp_fusion/o3d_utility.py
Connections 18
- Localization runner → Laser sensor model: scores scans
particle_filter/code/main.py - Localization runner → Particle resampling: resamples particles
particle_filter/code/main.py - Localization runner → Motion model: uses motion model
particle_filter/code/main.py - Localization runner → Occupancy map: loads map
particle_filter/code/main.py - Laser sensor model → Occupancy map: reads occupancy
particle_filter/code/sensor_model.py - Linear SLAM → Sparse solver methods: solves system
slam_solvers/code/linear.py - Linear SLAM → State and plotting utilities: uses state utilities
slam_solvers/code/linear.py - Nonlinear SLAM → Sparse solver methods: selects solver
slam_solvers/code/nonlinear.py - Nonlinear SLAM → State and plotting utilities: uses state utilities
slam_solvers/code/nonlinear.py - RGB-D mapping runner → Pose and normal preprocessing: loads poses
icp_fusion/main.py - RGB-D mapping runner → Camera transforms: unprojects depth
icp_fusion/main.py - RGB-D mapping runner → ICP alignment: aligns frames
icp_fusion/main.py - RGB-D mapping runner → Point-based fusion: fuses frame
icp_fusion/main.py - RGB-D mapping runner → Point-cloud utilities: builds visualization
icp_fusion/main.py - Pose and normal preprocessing → Camera transforms: unprojects depth
icp_fusion/preprocess.py - Pose and normal preprocessing → Point-cloud utilities: estimates normals
icp_fusion/preprocess.py - Point-based fusion → Camera transforms: projects points
icp_fusion/fusion.py - ICP alignment → Point-cloud utilities: visualizes matches
icp_fusion/icp.py