distributed_representation_of_APF/fig_influence_range.py at main · songfei-psy/distributed_representation_of_APF · GitHub

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
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
import numpy as np
import os
from Tools.plotting import Plotting_vector_map


class SquareMap:
    def __init__(self, N_sample, ratio=1):
        l = 50*ratio
        env_size = [l, l]
        self.x_range = [0, env_size[0]]
        self.y_range = [0, env_size[1]]
        self.goal = [env_size[0], env_size[1]]
        self.obs_point = [[env_size[0]/4, env_size[1]/4],[env_size[0]*3/4, env_size[1]*3/4]]
        self.point_r = 1*ratio

        x_sample = np.linspace(self.x_range[0], self.x_range[1], N_sample)
        y_sample = np.linspace(self.y_range[0], self.y_range[1], N_sample)
        X, Y = np.meshgrid(x_sample, y_sample)
        xx, yy = X.flatten(), Y.flatten()
        self.xx, self.yy = xx, yy


class APF:
    def __init__(self, l0, env):
        self.k_att = 1
        self.k_rep = 3000
        self.influence_radius = l0
        self.goal = env.goal
        self.env = env

    def attract(self, position):
        vector_to_goal = self.goal - position
        # distance_to_goal = np.linalg.norm(vector_to_goal)
        force = self.k_att * vector_to_goal
        return force

    def repulse(self, position):
        repulsive_forces = np.array([0.0, 0.0])
        for ox, oy in self.env.obs_point:
            obs_positon = np.array([ox, oy])
            repulsive_force, inner = self.calc_repulsive_force(position, obs_positon, self.env.point_r)
            if inner:
                return np.array([np.nan, np.nan]), True
            repulsive_forces += repulsive_force
        return repulsive_forces, False

    def calc_repulsive_force(self, position, obstacle_pos, obstacle_radius):
        vector_to_obstacle = position - obstacle_pos
        distance_to_obstacle = np.linalg.norm(vector_to_obstacle) - obstacle_radius
        if distance_to_obstacle >= self.influence_radius:
            return np.array([0.0, 0.0]), False
        if distance_to_obstacle < 0:
            return np.array([np.nan, np.nan]), True
        force = self.k_rep * (1.0 / distance_to_obstacle - 1.0 / self.influence_radius) \
                * (1.0 / distance_to_obstacle ** 2) * (vector_to_obstacle / np.linalg.norm(vector_to_obstacle))
        return force, False

    def decoding(self, position):
        F_rep, inner = self.repulse(position)
        if inner:
            return np.array([np.nan, np.nan])

        F_att = self.attract(position)
        force = F_att + F_rep
        return force / np.linalg.norm(force)

class baseDirection(APF):
    def __init__(self, l0, env):
        super().__init__(l0, env)

    def decoding(self, position):
        F_rep, inner = self.repulse(position)
        if inner:
            return np.array([np.nan, np.nan])
        F_att = self.attract(position)
        force = F_att
        return force / np.linalg.norm(force)


class NRF:
    def __init__(self, l0, env):
        self.kappa = 10
        self.influence_radius = l0
        self.pref_direction = np.linspace(0, 2 * np.pi, 100, endpoint=False)
        self.goal = env.goal
        self.env = env

    def attract(self, position):
        vector_to_goal = self.goal - position
        goal_dir = np.arctan2(vector_to_goal[1], vector_to_goal[0])

        return self.population(goal_dir, 1.5)

    def repulse(self, position):
        f_obs, L2 = [], []
        for ox, oy in self.env.obs_point:
            obs_positon = np.array([ox, oy])
            f, l, inner = self.calc_repulsive_force(position, obs_positon, self.env.point_r)
            if inner:
                return np.array([np.nan, np.nan]), True
            if f is not None:
                f_obs.append(f)
                L2.append(l)
        return self.norm_repulsive(L2, f_obs), False

    def calc_repulsive_force(self, position, obstacle_pos, obstacle_radius):
        vector_to_obstacle = position - obstacle_pos
        distance_to_obstacle = np.linalg.norm(vector_to_obstacle) - obstacle_radius
        if distance_to_obstacle > self.influence_radius:
            return None, None, False
        if distance_to_obstacle < 0:
            return None, None, True

        obs_dir = np.arctan2(vector_to_obstacle[1], vector_to_obstacle[0])
        nearst_obs = distance_to_obstacle
        ratio = (1.0 / nearst_obs - 1.0 / self.influence_radius) / (nearst_obs ** 2)
        L2 = np.exp(1j * obs_dir) * ratio
        f_obs = self.population(obs_dir, self.kappa) * ratio
        return f_obs, L2, False

    def population(self, direction, kappa):
        return np.exp(kappa * np.cos(direction - self.pref_direction)) / np.exp(kappa)

    def decoding(self, position):
        F_rep, inner = self.repulse(position)
        if inner:
            return np.array([np.nan, np.nan])

        F_att = self.attract(position)
        F_sum = F_att + F_rep
        FF = np.sum(np.exp(1j * self.pref_direction) * F_sum)
        current_direction = np.angle(FF)
        force = np.array([np.cos(current_direction), np.sin(current_direction)])
        return force

    def norm_repulsive(self, L2, f_obs):
        L2_norm = np.abs(np.sum(np.array(L2)))
        if L2_norm > 0:
            f_obs = np.array(f_obs)
            F_obs = f_obs / L2_norm
            F_obs = np.sum(F_obs, 0)
        else:
            F_obs = np.zeros_like(self.pref_direction)
        return F_obs


if __name__ == '__main__':
    # folder
    project_folder = os.path.dirname(os.path.abspath(__file__))
    data_folder = os.path.join(project_folder, 'data', 'method')
    if not os.path.exists(data_folder):
        os.makedirs(data_folder)
    figure_appendix = os.path.join(data_folder, f'influence_range_')

    ratios = [1, 2]
    influence_range = [12]
    N_sample = 25
    vector_map_apf, vector_map_nrf, vector_map_base = [], [], []
    l0, envs = [], []
    for ratio in ratios:
        env = SquareMap(N_sample, ratio)
        xx, yy = env.xx, env.yy

        # planning
        for influence_rad in influence_range:
            vector_apf, vector_nrf, vector_base = [], [], []
            nrf = NRF(influence_rad*ratio, env)
            apf = APF(influence_rad*ratio, env)
            base = baseDirection(influence_rad*ratio, env)
            for x, y in zip(xx, yy):
                position = np.array([x, y])
                vector_apf.append(apf.decoding(position))
                vector_nrf.append(nrf.decoding(position))
                vector_base.append(base.decoding(position))

            vector_map_apf.append(vector_apf)
            vector_map_nrf.append(vector_nrf)
            vector_map_base.append(vector_base)
            l0.append(influence_rad*ratio)
            envs.append(env)

    # plot
    plotting = Plotting_vector_map(l0, envs, figure_appendix)
    plotting.showing(vector_map_apf, vector_map_nrf, vector_map_base)