Strategic Objectives
• Master the mathematical foundations of factor graph representations.
• Implement robust back-end optimization techniques for large-scale maps.
• Understand the mechanics of pose-graph relaxation and loop closure.
• Bridge the gap between raw spatial data and global trajectory consistency.
The Core Challenge
Robots often lose their way as sensor errors accumulate, turning precise paths into a tangled web of uncertainty.
01
The Back-End Revolution
02
The Geometry of Space
03
Foundations of Factor Graphs
04
The Probabilistic Backbone
05
Non-Linear Least Squares
06
The Gauss-Newton Method
07
The Levenberg-Marquardt Logic
08
Sparsity and the Jacobian
09
Lie Groups in Robotics
10
Pose-Graph Relaxation
11
Loop Closure Detection
12
The Information Matrix
13
Maximum Likelihood Estimation
14
Robust Kernels
15
Marginalization and Sparsification
16
Incremental Smoothing
17
Visual-Inertial Odometry Graphs
18
The Schur Complement
19
Hypergraph Architectures
20
Error Analysis and Metrics
21