1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
|
#include "kalman_filter.hpp"
void kalman_filter::matrix_initialize() {
state_vector.setZero(n);
state_covariance.setZero(n, n);
state_transition_M = MatrixXf::Zero(n, n);
control_input_M = MatrixXf::Zero(n, p);
I = MatrixXf::Identity(n, n);
measurement_M.setIdentity(m, n); // Setup Measurement Matrix
process_noise_covariance = MatrixXf::Zero(n, n);
measurement_covariance = MatrixXf::Zero(m, m);
// Setup State Transition Matrix
state_transition_M << 1.0, dt, 0.0, 1.0;
// Setup Control Input Matrix
control_input_M << 0.5 * dt * dt, dt; // (Linear Displacement Eq.)
// Setup Process-Noise Covariance
process_noise_covariance(0,0) = 0.01;
process_noise_covariance(1,1) = 0.1;
// Setup Measurement Covariance
measurement_covariance << 1e-12;
}
void kalman_filter::matrix_update() {
state_transition_M(0, 1) = dt;
control_input_M(0, 0) = 0.5f * dt * dt;
control_input_M(1, 0) = dt;
}
void kalman_filter::predict(VectorXf control_vec) {
state_vector = (state_transition_M * state_vector) + (control_input_M * control_vec);
state_covariance = (state_transition_M * (state_covariance * state_transition_M.transpose())) + process_noise_covariance;
}
void kalman_filter::update(VectorXf measurement) {
// Innovation
VectorXf y = measurement - (measurement_M * state_vector);
// Residual/Innovation Covariance
MatrixXf S = (measurement_M * (state_covariance * measurement_M.transpose())) + measurement_covariance;
// Kalman Gain
MatrixXf K = (state_covariance * measurement_M.transpose()) * S.inverse();
// Update
state_vector = state_vector + (K * y);
state_covariance = (I - (K * measurement_M)) * state_covariance;
}
kalman_filter::kalman_filter(int state_dim, int control_dim, int measurement_dim, float dt) : n(state_dim), p(control_dim), m(measurement_dim), dt(dt) {
matrix_initialize();
}
bool kalman_filter::state_initialize(VectorXf state_vec, MatrixXf state_cov) {
bool result { false };
if (state_vec.size() == n && state_cov.rows() == n) {
state_vector = state_vec;
state_covariance = state_cov;
result = true;
}
return result;
}
VectorXf kalman_filter::run(VectorXf control, VectorXf measurement, float _dt) {
if (control.size() == p && measurement.size() == m) {
dt = _dt;
matrix_update();
predict(control);
update(measurement);
}
return state_vector;
}
|
