summaryrefslogtreecommitdiffstats
path: root/ml/dlib/dlib/test/filtering.cpp
blob: 61dc88440373eda959749de9b7ad902718025a7c (plain)
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
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
// Copyright (C) 2012  Davis E. King (davis@dlib.net)
// License: Boost Software License   See LICENSE.txt for the full license.


#include <dlib/filtering.h>
#include <sstream>
#include <string>
#include <cstdlib>
#include <ctime>
#include <dlib/matrix.h>
#include <dlib/rand.h>

#include "tester.h"

namespace  
{

    using namespace test;
    using namespace dlib;
    using namespace std;

    logger dlog("test.filtering");

// ----------------------------------------------------------------------------------------

    template <typename filter_type>
    double test_filter (
        filter_type kf,
        int size
    )
    {
        // This test has a point moving in a circle around the origin.  The point
        // also gets a random bump in a random direction at each time step.

        running_stats<double> rs;

        dlib::rand rnd;
        int count = 0;
        const dlib::vector<double,3> z(0,0,1);
        dlib::vector<double,2> p(10,10), temp;
        for (int i = 0; i < size; ++i)
        {
            // move the point around in a circle
            p += z.cross(p).normalize()/0.5;
            // randomly drop measurements
            if (rnd.get_random_double() < 0.7 || count < 4)
            {
                // make a random bump
                dlib::vector<double,2> pp;
                pp.x() = rnd.get_random_gaussian()/3;
                pp.y() = rnd.get_random_gaussian()/3;

                ++count;
                kf.update(p+pp);
            }
            else
            {
                kf.update();
                dlog << LTRACE << "MISSED MEASUREMENT";
            }
            // figure out the next position
            temp = (p+z.cross(p).normalize()/0.5);
            const double error = length(temp - rowm(kf.get_predicted_next_state(),range(0,1)));
            rs.add(error);

            dlog << LTRACE << temp << "("<< error << "): " << trans(kf.get_predicted_next_state());

            // test the serialization a few times.
            if (count < 10)
            {
                ostringstream sout;
                serialize(kf, sout);
                istringstream sin(sout.str());
                filter_type temp;
                deserialize(temp, sin);
                kf = temp;
            }
        }


        return rs.mean();

    }

// ----------------------------------------------------------------------------------------

    void test_kalman_filter()
    {
        matrix<double,2,2> R;
        R = 0.3, 0,
        0,  0.3;

        // the variables in the state are 
        // x,y, x velocity, y velocity, x acceleration, and y acceleration
        matrix<double,6,6> A;
        A = 1, 0, 1, 0, 0, 0,
        0, 1, 0, 1, 0, 0,
        0, 0, 1, 0, 1, 0,
        0, 0, 0, 1, 0, 1,
        0, 0, 0, 0, 1, 0,
        0, 0, 0, 0, 0, 1;

        // the measurements only tell us the positions
        matrix<double,2,6> H;
        H = 1, 0, 0, 0, 0, 0,
        0, 1, 0, 0, 0, 0;


        kalman_filter<6,2> kf; 
        kf.set_measurement_noise(R);  
        matrix<double> pn = 0.01*identity_matrix<double,6>();
        kf.set_process_noise(pn);
        kf.set_observation_model(H);
        kf.set_transition_model(A);

        DLIB_TEST(equal(kf.get_observation_model() , H));
        DLIB_TEST(equal(kf.get_transition_model() , A));
        DLIB_TEST(equal(kf.get_measurement_noise() , R));
        DLIB_TEST(equal(kf.get_process_noise() , pn));
        DLIB_TEST(equal(kf.get_current_estimation_error_covariance() , identity_matrix(pn)));

        double kf_error = test_filter(kf, 300);

        dlog << LINFO << "kf error: "<< kf_error;
        DLIB_TEST_MSG(kf_error < 0.75, kf_error);
    }

// ----------------------------------------------------------------------------------------

    void test_rls_filter()
    {

        rls_filter rls(10, 0.99, 0.1);

        DLIB_TEST(rls.get_window_size() == 10);
        DLIB_TEST(rls.get_forget_factor() == 0.99);
        DLIB_TEST(rls.get_c() == 0.1);

        double rls_error = test_filter(rls, 1000);

        dlog << LINFO << "rls error: "<< rls_error;
        DLIB_TEST_MSG(rls_error < 0.75, rls_error);
    }

// ----------------------------------------------------------------------------------------

    class filtering_tester : public tester
    {
    public:
        filtering_tester (
        ) :
            tester ("test_filtering",
                    "Runs tests on the filtering stuff (rls and kalman filters).")
        {}

        void perform_test (
        )
        {
            test_rls_filter();
            test_kalman_filter();
        }
    } a;

}