Trajectron-plus-plus/experiments/nuScenes/process_data.py
2020-04-05 21:43:49 -04:00

472 lines
No EOL
20 KiB
Python

import sys
import os
import numpy as np
import pandas as pd
import dill
import argparse
from tqdm import tqdm
from pyquaternion import Quaternion
from kalman_filter import NonlinearKinematicBicycle
from sklearn.model_selection import train_test_split
nu_path = './devkit/python-sdk/'
sys.path.append(nu_path)
sys.path.append("../../trajectron")
from nuscenes.nuscenes import NuScenes
from nuscenes.map_expansion.map_api import NuScenesMap
from nuscenes.utils.splits import create_splits_scenes
from environment import Environment, Scene, Node, GeometricMap, derivative_of
scene_blacklist = [499, 515, 517]
FREQUENCY = 2
dt = 1 / FREQUENCY
data_columns_vehicle = pd.MultiIndex.from_product([['position', 'velocity', 'acceleration', 'heading'], ['x', 'y']])
data_columns_vehicle = data_columns_vehicle.append(pd.MultiIndex.from_tuples([('heading', '°'), ('heading', '')]))
data_columns_vehicle = data_columns_vehicle.append(pd.MultiIndex.from_product([['velocity', 'acceleration'], ['norm']]))
data_columns_pedestrian = pd.MultiIndex.from_product([['position', 'velocity', 'acceleration'], ['x', 'y']])
curv_0_2 = 0
curv_0_1 = 0
total = 0
standardization = {
'PEDESTRIAN': {
'position': {
'x': {'mean': 0, 'std': 1},
'y': {'mean': 0, 'std': 1}
},
'velocity': {
'x': {'mean': 0, 'std': 2},
'y': {'mean': 0, 'std': 2}
},
'acceleration': {
'x': {'mean': 0, 'std': 1},
'y': {'mean': 0, 'std': 1}
}
},
'VEHICLE': {
'position': {
'x': {'mean': 0, 'std': 80},
'y': {'mean': 0, 'std': 80}
},
'velocity': {
'x': {'mean': 0, 'std': 15},
'y': {'mean': 0, 'std': 15},
'norm': {'mean': 0, 'std': 15}
},
'acceleration': {
'x': {'mean': 0, 'std': 4},
'y': {'mean': 0, 'std': 4},
'norm': {'mean': 0, 'std': 4}
},
'heading': {
'x': {'mean': 0, 'std': 1},
'y': {'mean': 0, 'std': 1},
'°': {'mean': 0, 'std': np.pi},
'': {'mean': 0, 'std': 1}
}
}
}
def augment_scene(scene, angle):
def rotate_pc(pc, alpha):
M = np.array([[np.cos(alpha), -np.sin(alpha)],
[np.sin(alpha), np.cos(alpha)]])
return M @ pc
data_columns_vehicle = pd.MultiIndex.from_product([['position', 'velocity', 'acceleration', 'heading'], ['x', 'y']])
data_columns_vehicle = data_columns_vehicle.append(pd.MultiIndex.from_tuples([('heading', '°'), ('heading', '')]))
data_columns_vehicle = data_columns_vehicle.append(pd.MultiIndex.from_product([['velocity', 'acceleration'], ['norm']]))
data_columns_pedestrian = pd.MultiIndex.from_product([['position', 'velocity', 'acceleration'], ['x', 'y']])
scene_aug = Scene(timesteps=scene.timesteps, dt=scene.dt, name=scene.name, non_aug_scene=scene)
alpha = angle * np.pi / 180
for node in scene.nodes:
if node.type == 'PEDESTRIAN':
x = node.data.position.x.copy()
y = node.data.position.y.copy()
x, y = rotate_pc(np.array([x, y]), alpha)
vx = derivative_of(x, scene.dt)
vy = derivative_of(y, scene.dt)
ax = derivative_of(vx, scene.dt)
ay = derivative_of(vy, scene.dt)
data_dict = {('position', 'x'): x,
('position', 'y'): y,
('velocity', 'x'): vx,
('velocity', 'y'): vy,
('acceleration', 'x'): ax,
('acceleration', 'y'): ay}
node_data = pd.DataFrame(data_dict, columns=data_columns_pedestrian)
node = Node(node_type=node.type, node_id=node.id, data=node_data, first_timestep=node.first_timestep)
elif node.type == 'VEHICLE':
x = node.data.position.x.copy()
y = node.data.position.y.copy()
heading = getattr(node.data.heading, '°').copy()
heading += alpha
heading = (heading + np.pi) % (2.0 * np.pi) - np.pi
x, y = rotate_pc(np.array([x, y]), alpha)
vx = derivative_of(x, scene.dt)
vy = derivative_of(y, scene.dt)
ax = derivative_of(vx, scene.dt)
ay = derivative_of(vy, scene.dt)
v = np.stack((vx, vy), axis=-1)
v_norm = np.linalg.norm(np.stack((vx, vy), axis=-1), axis=-1, keepdims=True)
heading_v = np.divide(v, v_norm, out=np.zeros_like(v), where=(v_norm > 1.))
heading_x = heading_v[:, 0]
heading_y = heading_v[:, 1]
data_dict = {('position', 'x'): x,
('position', 'y'): y,
('velocity', 'x'): vx,
('velocity', 'y'): vy,
('velocity', 'norm'): np.linalg.norm(np.stack((vx, vy), axis=-1), axis=-1),
('acceleration', 'x'): ax,
('acceleration', 'y'): ay,
('acceleration', 'norm'): np.linalg.norm(np.stack((ax, ay), axis=-1), axis=-1),
('heading', 'x'): heading_x,
('heading', 'y'): heading_y,
('heading', '°'): heading,
('heading', ''): derivative_of(heading, dt, radian=True)}
node_data = pd.DataFrame(data_dict, columns=data_columns_vehicle)
node = Node(node_type=node.type, node_id=node.id, data=node_data, first_timestep=node.first_timestep,
non_aug_node=node)
scene_aug.nodes.append(node)
return scene_aug
def augment(scene):
scene_aug = np.random.choice(scene.augmented)
scene_aug.temporal_scene_graph = scene.temporal_scene_graph
scene_aug.map = scene.map
return scene_aug
def trajectory_curvature(t):
path_distance = np.linalg.norm(t[-1] - t[0])
lengths = np.sqrt(np.sum(np.diff(t, axis=0) ** 2, axis=1)) # Length between points
path_length = np.sum(lengths)
if np.isclose(path_distance, 0.):
return 0, 0, 0
return (path_length / path_distance) - 1, path_length, path_distance
def process_scene(ns_scene, env, nusc, data_path):
scene_id = int(ns_scene['name'].replace('scene-', ''))
data = pd.DataFrame(columns=['frame_id',
'type',
'node_id',
'robot',
'x', 'y', 'z',
'length',
'width',
'height',
'heading'])
sample_token = ns_scene['first_sample_token']
sample = nusc.get('sample', sample_token)
frame_id = 0
while sample['next']:
annotation_tokens = sample['anns']
for annotation_token in annotation_tokens:
annotation = nusc.get('sample_annotation', annotation_token)
category = annotation['category_name']
if len(annotation['attribute_tokens']):
attribute = nusc.get('attribute', annotation['attribute_tokens'][0])['name']
else:
continue
if 'pedestrian' in category and not 'stroller' in category and not 'wheelchair' in category:
our_category = env.NodeType.PEDESTRIAN
elif 'vehicle' in category and 'bicycle' not in category and 'motorcycle' not in category and 'parked' not in attribute:
our_category = env.NodeType.VEHICLE
else:
continue
data_point = pd.Series({'frame_id': frame_id,
'type': our_category,
'node_id': annotation['instance_token'],
'robot': False,
'x': annotation['translation'][0],
'y': annotation['translation'][1],
'z': annotation['translation'][2],
'length': annotation['size'][0],
'width': annotation['size'][1],
'height': annotation['size'][2],
'heading': Quaternion(annotation['rotation']).yaw_pitch_roll[0]})
data = data.append(data_point, ignore_index=True)
# Ego Vehicle
our_category = env.NodeType.VEHICLE
sample_data = nusc.get('sample_data', sample['data']['CAM_FRONT'])
annotation = nusc.get('ego_pose', sample_data['ego_pose_token'])
data_point = pd.Series({'frame_id': frame_id,
'type': our_category,
'node_id': 'ego',
'robot': True,
'x': annotation['translation'][0],
'y': annotation['translation'][1],
'z': annotation['translation'][2],
'length': 4,
'width': 1.7,
'height': 1.5,
'heading': Quaternion(annotation['rotation']).yaw_pitch_roll[0],
'orientation': None})
data = data.append(data_point, ignore_index=True)
sample = nusc.get('sample', sample['next'])
frame_id += 1
if len(data.index) == 0:
return None
data.sort_values('frame_id', inplace=True)
max_timesteps = data['frame_id'].max()
x_min = np.round(data['x'].min() - 50)
x_max = np.round(data['x'].max() + 50)
y_min = np.round(data['y'].min() - 50)
y_max = np.round(data['y'].max() + 50)
data['x'] = data['x'] - x_min
data['y'] = data['y'] - y_min
scene = Scene(timesteps=max_timesteps + 1, dt=dt, name=str(scene_id), aug_func=augment)
# Generate Maps
map_name = nusc.get('log', ns_scene['log_token'])['location']
nusc_map = NuScenesMap(dataroot=data_path, map_name=map_name)
type_map = dict()
x_size = x_max - x_min
y_size = y_max - y_min
patch_box = (x_min + 0.5 * (x_max - x_min), y_min + 0.5 * (y_max - y_min), y_size, x_size)
patch_angle = 0 # Default orientation where North is up
canvas_size = (np.round(3 * y_size).astype(int), np.round(3 * x_size).astype(int))
homography = np.array([[3., 0., 0.], [0., 3., 0.], [0., 0., 3.]])
layer_names = ['lane', 'road_segment', 'drivable_area', 'road_divider', 'lane_divider', 'stop_line',
'ped_crossing', 'stop_line', 'ped_crossing', 'walkway']
map_mask = (nusc_map.get_map_mask(patch_box, patch_angle, layer_names, canvas_size) * 255.0).astype(
np.uint8)
map_mask = np.swapaxes(map_mask, 1, 2) # x axis comes first
# PEDESTRIANS
map_mask_pedestrian = np.stack((map_mask[9], map_mask[8], np.max(map_mask[:3], axis=0)), axis=0)
type_map['PEDESTRIAN'] = GeometricMap(data=map_mask_pedestrian, homography=homography, description=', '.join(layer_names))
# VEHICLES
map_mask_vehicle = np.stack((np.max(map_mask[:3], axis=0), map_mask[3], map_mask[4]), axis=0)
type_map['VEHICLE'] = GeometricMap(data=map_mask_vehicle, homography=homography, description=', '.join(layer_names))
map_mask_plot = np.stack(((np.max(map_mask[:3], axis=0) - (map_mask[3] + 0.5 * map_mask[4]).clip(
max=255)).clip(min=0).astype(np.uint8), map_mask[8], map_mask[9]), axis=0)
type_map['VISUALIZATION'] = GeometricMap(data=map_mask_plot, homography=homography, description=', '.join(layer_names))
scene.map = type_map
del map_mask
del map_mask_pedestrian
del map_mask_vehicle
del map_mask_plot
for node_id in pd.unique(data['node_id']):
node_frequency_multiplier = 1
node_df = data[data['node_id'] == node_id]
if node_df['x'].shape[0] < 2:
continue
if not np.all(np.diff(node_df['frame_id']) == 1):
# print('Occlusion')
continue # TODO Make better
node_values = node_df[['x', 'y']].values
x = node_values[:, 0]
y = node_values[:, 1]
heading = node_df['heading'].values
if node_df.iloc[0]['type'] == env.NodeType.VEHICLE and not node_id == 'ego':
# Kalman filter Agent
vx = derivative_of(x, scene.dt)
vy = derivative_of(y, scene.dt)
velocity = np.linalg.norm(np.stack((vx, vy), axis=-1), axis=-1)
filter_veh = NonlinearKinematicBicycle(dt=scene.dt, sMeasurement=1.0)
P_matrix = None
for i in range(len(x)):
if i == 0: # initalize KF
# initial P_matrix
P_matrix = np.identity(4)
elif i < len(x):
# assign new est values
x[i] = x_vec_est_new[0][0]
y[i] = x_vec_est_new[1][0]
heading[i] = x_vec_est_new[2][0]
velocity[i] = x_vec_est_new[3][0]
if i < len(x) - 1: # no action on last data
# filtering
x_vec_est = np.array([[x[i]],
[y[i]],
[heading[i]],
[velocity[i]]])
z_new = np.array([[x[i + 1]],
[y[i + 1]],
[heading[i + 1]],
[velocity[i + 1]]])
x_vec_est_new, P_matrix_new = filter_veh.predict_and_update(
x_vec_est=x_vec_est,
u_vec=np.array([[0.], [0.]]),
P_matrix=P_matrix,
z_new=z_new
)
P_matrix = P_matrix_new
curvature, pl, _ = trajectory_curvature(np.stack((x, y), axis=-1))
if pl < 1.0: # vehicle is "not" moving
x = x[0].repeat(max_timesteps + 1)
y = y[0].repeat(max_timesteps + 1)
heading = heading[0].repeat(max_timesteps + 1)
global total
global curv_0_2
global curv_0_1
total += 1
if pl > 1.0:
if curvature > .2:
curv_0_2 += 1
node_frequency_multiplier = 3*int(np.floor(total/curv_0_2))
elif curvature > .1:
curv_0_1 += 1
node_frequency_multiplier = 3*int(np.floor(total/curv_0_1))
vx = derivative_of(x, scene.dt)
vy = derivative_of(y, scene.dt)
ax = derivative_of(vx, scene.dt)
ay = derivative_of(vy, scene.dt)
if node_df.iloc[0]['type'] == env.NodeType.VEHICLE:
v = np.stack((vx, vy), axis=-1)
v_norm = np.linalg.norm(np.stack((vx, vy), axis=-1), axis=-1, keepdims=True)
heading_v = np.divide(v, v_norm, out=np.zeros_like(v), where=(v_norm > 1.))
heading_x = heading_v[:, 0]
heading_y = heading_v[:, 1]
data_dict = {('position', 'x'): x,
('position', 'y'): y,
('velocity', 'x'): vx,
('velocity', 'y'): vy,
('velocity', 'norm'): np.linalg.norm(np.stack((vx, vy), axis=-1), axis=-1),
('acceleration', 'x'): ax,
('acceleration', 'y'): ay,
('acceleration', 'norm'): np.linalg.norm(np.stack((ax, ay), axis=-1), axis=-1),
('heading', 'x'): heading_x,
('heading', 'y'): heading_y,
('heading', '°'): heading,
('heading', ''): derivative_of(heading, dt, radian=True)}
node_data = pd.DataFrame(data_dict, columns=data_columns_vehicle)
else:
data_dict = {('position', 'x'): x,
('position', 'y'): y,
('velocity', 'x'): vx,
('velocity', 'y'): vy,
('acceleration', 'x'): ax,
('acceleration', 'y'): ay}
node_data = pd.DataFrame(data_dict, columns=data_columns_pedestrian)
node = Node(node_type=node_df.iloc[0]['type'], node_id=node_id, data=node_data, frequency_multiplier=node_frequency_multiplier)
node.first_timestep = node_df['frame_id'].iloc[0]
if node_df.iloc[0]['robot'] == True:
node.is_robot = True
scene.robot = node
scene.nodes.append(node)
return scene
def process_data(data_path, version, output_path, val_split):
nusc = NuScenes(version=version, dataroot=data_path, verbose=True)
splits = create_splits_scenes()
train_scenes, val_scenes = train_test_split(splits['train' if 'mini' not in version else 'mini_train'], test_size=val_split)
train_scene_names = splits['train' if 'mini' not in version else 'mini_train']
val_scene_names = []#val_scenes
test_scene_names = splits['val' if 'mini' not in version else 'mini_val']
ns_scene_names = dict()
ns_scene_names['train'] = train_scene_names
ns_scene_names['val'] = val_scene_names
ns_scene_names['test'] = test_scene_names
for data_class in ['train', 'val', 'test']:
env = Environment(node_type_list=['VEHICLE', 'PEDESTRIAN'], standardization=standardization)
attention_radius = dict()
attention_radius[(env.NodeType.PEDESTRIAN, env.NodeType.PEDESTRIAN)] = 10.0
attention_radius[(env.NodeType.PEDESTRIAN, env.NodeType.VEHICLE)] = 20.0
attention_radius[(env.NodeType.VEHICLE, env.NodeType.PEDESTRIAN)] = 20.0
attention_radius[(env.NodeType.VEHICLE, env.NodeType.VEHICLE)] = 30.0
env.attention_radius = attention_radius
env.robot_type = env.NodeType.VEHICLE
scenes = []
for ns_scene_name in tqdm(ns_scene_names[data_class]):
ns_scene = nusc.get('scene', nusc.field2token('scene', 'name', ns_scene_name)[0])
scene_id = int(ns_scene['name'].replace('scene-', ''))
if scene_id in scene_blacklist: # Some scenes have bad localization
continue
scene = process_scene(ns_scene, env, nusc, data_path)
if scene is not None:
if data_class == 'train':
scene.augmented = list()
angles = np.arange(0, 360, 15)
for angle in angles:
scene.augmented.append(augment_scene(scene, angle))
scenes.append(scene)
print(f'Processed {len(scenes):.2f} scenes')
env.scenes = scenes
if len(scenes) > 0:
mini_string = ''
if 'mini' in version:
mini_string = '_mini'
data_dict_path = os.path.join(output_path, 'nuScenes_' + data_class + mini_string + '_full.pkl')
with open(data_dict_path, 'wb') as f:
dill.dump(env, f, protocol=dill.HIGHEST_PROTOCOL)
print('Saved Environment!')
global total
global curv_0_2
global curv_0_1
print(f"Total Nodes: {total}")
print(f"Curvature > 0.1 Nodes: {curv_0_1}")
print(f"Curvature > 0.2 Nodes: {curv_0_2}")
total = 0
curv_0_1 = 0
curv_0_2 = 0
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--data', type=str, required=True)
parser.add_argument('--version', type=str, required=True)
parser.add_argument('--output_path', type=str, required=True)
parser.add_argument('--val_split', type=int, default=0.15)
args = parser.parse_args()
process_data(args.data, args.version, args.output_path, args.val_split)