ga-based_mapfusion/map_generator.py at main · cristinaluna/ga-based_mapfusion · 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
'''

Map generator


Authors: C. Luna, E. Ruiz, P. López, D. F. Barrero
Maintainer: @cristinaluna

'''


import numpy as np
import cv2 as cv
import random
from scipy.ndimage import rotate


def draw_room(img, x, y, w, h, filled=True):
    shape_type = random.choice(['rect', 'rotated_rect', 'irregular'])
    if shape_type == 'rect':
        cv.rectangle(img, (x, y), (x + w, y + h), 255, -1 if filled else 2)
    elif shape_type == 'rotated_rect':
        angle = random.uniform(-30, 30)
        rect = ((x + w // 2, y + h // 2), (w, h), angle)
        box = cv.boxPoints(rect).astype(int)
        cv.drawContours(img, [box], 0, 255, -1)
    elif shape_type == 'irregular':
        pts = np.array([
            [x, y],
            [x + w, y + random.randint(-10, 10)],
            [x + w + random.randint(-10, 10), y + h],
            [x + random.randint(-10, 10), y + h + random.randint(-10, 10)]
        ])
        cv.fillPoly(img, [pts], 255)

    # Larger obstacles (more and bigger)
    for _ in range(random.randint(5, 10)):
        ox = x + random.randint(10, max(10, w - 40))
        oy = y + random.randint(10, max(10, h - 40))
        ow = random.randint(15, min(30, w - (ox - x)))
        oh = random.randint(15, min(30, h - (oy - y)))
        cv.rectangle(img, (ox, oy), (ox + ow, oy + oh), 0, -1)


def connect_rooms_with_corridor(img, room1, room2):
    x1, y1, w1, h1 = room1
    x2, y2, w2, h2 = room2

    edges1 = [
        (x1 + w1 // 2, y1),           # top center
        (x1 + w1 // 2, y1 + h1),      # bottom center
        (x1, y1 + h1 // 2),           # left center
        (x1 + w1, y1 + h1 // 2)       # right center
    ]
    edges2 = [
        (x2 + w2 // 2, y2),
        (x2 + w2 // 2, y2 + h2),
        (x2, y2 + h2 // 2),
        (x2 + w2, y2 + h2 // 2)
    ]

    min_dist = float('inf')
    best_pair = (edges1[0], edges2[0])
    for e1 in edges1:
        for e2 in edges2:
            dist = (e1[0] - e2[0]) ** 2 + (e1[1] - e2[1]) ** 2
            if dist < min_dist:
                min_dist = dist
                best_pair = (e1, e2)

    (x_start, y_start), (x_end, y_end) = best_pair

    thickness = random.randint(15, 25)

    if random.random() > 0.5:
        mid_point = (x_end, y_start)
    else:
        mid_point = (x_start, y_end)

    cv.line(img, (x_start, y_start), mid_point, 255, thickness)
    cv.line(img, mid_point, (x_end, y_end), 255, thickness)

    for _ in range(random.randint(6, 10)):
        tx = random.randint(min(x_start, x_end), max(x_start, x_end))
        ty = random.randint(min(y_start, y_end), max(y_start, y_end))
        ow = random.randint(10, 20)
        oh = random.randint(10, 20)
        cv.rectangle(img, (tx, ty), (tx + ow, ty + oh), 0, -1)


def generate_overlapping_submaps(global_size=(1200, 1200), submap_size=(600, 600),
                                 overlap_ratio=0.6, num_rooms=6, seed=None,
                                 apply_rotation=True, max_rotation=5,
                                 apply_noise=True, noise_std=0.02):
    if seed is not None:
        random.seed(seed)
        np.random.seed(seed)

    H, W = global_size
    sub_h, sub_w = submap_size

    base = np.zeros((H, W), dtype=np.uint8)
    rooms = []

    for _ in range(num_rooms):
        rw = random.randint(120, 180)  # bigger rooms
        rh = random.randint(120, 180)
        x = random.randint(0, W - rw - 1)
        y = random.randint(0, H - rh - 1)
        draw_room(base, x, y, rw, rh)
        rooms.append((x, y, rw, rh))

    for i in range(len(rooms) - 1):
        connect_rooms_with_corridor(base, rooms[i], rooms[i + 1])

    overlap_px_h = int(sub_h * overlap_ratio)
    overlap_px_w = int(sub_w * overlap_ratio)

    margin = 30
    max_y = H - sub_h - margin
    max_x = W - sub_w - margin

    y1 = random.randint(margin, max_y)
    x1 = random.randint(margin, max_x)

    y2 = y1 + sub_h - overlap_px_h + random.randint(-15, 15)
    x2 = x1 + sub_w - overlap_px_w + random.randint(-15, 15)

    y2 = np.clip(y2, margin, max_y)
    x2 = np.clip(x2, margin, max_x)

    submap1 = base[y1:y1 + sub_h, x1:x1 + sub_w].copy()
    submap2 = base[y2:y2 + sub_h, x2:x2 + sub_w].copy()

    if apply_rotation:
        angle = random.uniform(-max_rotation, max_rotation)  # subtle rotation
        submap2 = rotate(submap2, angle, reshape=False, order=1, mode='constant', cval=0)
        submap2 = (submap2 > 0.5).astype(np.uint8)

    if apply_noise:
        noise = np.random.normal(0, noise_std, submap2.shape)
        submap2 = np.clip(submap2.astype(np.float32) + noise, 0, 1)
        submap2 = (submap2 > 0.5).astype(np.uint8)

    return (submap1 > 0).astype(np.uint8), (submap2 > 0).astype(np.uint8)