commit c7f94cd2bd63a1f5201bbe1a1330396a4535be1e
parent 452549eb675e98eab51ee3b6a6aadd5eea811a0d
Author: Andrew Laack <andrew@laack.co>
Date: Sun, 19 Oct 2025 21:03:37 -0500
Took notes based on results of exam
Diffstat:
2 files changed, 8 insertions(+), 0 deletions(-)
diff --git a/docs/KalmanFilters.md b/docs/KalmanFilters.md
@@ -23,3 +23,10 @@ $\sigma^2_c = \frac{1}{\frac{1}{\sigma^2_a} + \frac{1}{\sigma^2_b}}$
Higher dimensional gaussians can useful for tracking both position and velocity to improve next step predictions.
Gaussian combination is easy in 1D. The generalization is a bit less so, and uses matrices.
+
+A few variables in use for KFs in higher dimensions are:
+
+- F : this matrix is the state transition matrix. This is multiplied by the state vector to predict positions at timestep t+1
+- H : This is the observation matrix.
+- X_t : This is the transpose of the state estimation vector.
+- P: Estimate uncertainty matrix.
diff --git a/docs/ParticleFilters.md b/docs/ParticleFilters.md
@@ -15,3 +15,4 @@
1. This is proportional to how likely the particle measurement is given the robot's measurement.
3. Sample particles proportional to their importance weight
1. This involves sampling particles with replacement to survive to the next step. This ensures we have a consistent number of particles from turn to turn, and that they are decent representations of the robot's true state.
+ 4. Fuzz (apply pertubations to surviving particles to induce diversity)
