运行NTU数据集

[black-magics]

我在使用的过程中能够跑出轨迹但是与实际轨迹偏差太大，提取的特征点只有30多个。我已经使用rosbag的方式成功运行了euroc数据集。

这是我配置的yaml文件
%YAML:1.0

image_height: 480
image_width: 752
use_imu: 1

depth_lower_thr: 0.05
depth_upper_thr: 20
max_y_diff: 2

Calibration

distortion_type: 1 # 0 for undistorted inputs, 1 for radial-tangential: [k1, k2, p1, p2, k3], 2 for equidistant/fisheye: [k1, k2, k3, k4, 0].

左目相机（cam0）

cam0:
intrinsics: [425.0258563372763, 426.7976260903337, 386.015186655088, 241.913033674344] # fx, fy, cx, cy
distortion_coeffs: [-0.288105327549552, 0.074578284234601, 0.0007784489598138802, -0.0002277853975035461, 0.0]
distortion_type: 1 # 1 表示 radial-tangential（默认）
T_type: 0 # 0: Tbc，IMU → cam0
T:
- [ 0.02183084, -0.01312053, 0.99967558, 0.00552943]
- [ 0.99975965, 0.00230088, -0.02180248, -0.12431302]
- [-0.00201407, 0.99991127, 0.01316761, 0.01614686]
- [ 0.0, 0.0, 0.0, 1.0 ]

右目相机（cam1）

cam1:
intrinsics: [431.3364265799752, 432.7527965378035, 354.8956286992647, 232.5508916495161] # fx, fy, cx, cy
distortion_coeffs: [-0.300267420221178, 0.090544063693053, 3.330220891093334e-05, 8.989607188457415e-05, 0.0]
distortion_type: 1
T_type: 0
T:
- [-0.01916508, -0.01496218, 0.99970437, 0.00519443]
- [ 0.99974371, 0.01176483, 0.01934191, 0.1347802 ]
- [-0.01205075, 0.99981884, 0.01473287, 0.01465067]
- [ 0.0, 0.0, 0.0, 1.0 ]

IMU

rate_hz: 200
gyroscope_noise_density: 1.6968e-04 # rad/s/√Hz
gyroscope_random_walk: 1.9393e-05 # rad/s²/√Hz
accelerometer_noise_density: 2.0000e-03 # m/s²/√Hz
accelerometer_random_walk: 3.0000e-03 # m/s³/√Hz
g_value: 9.81007
还有vo_ntu.yaml文件：
plnet:
use_superpoint: 1
max_keypoints: 1000
keypoint_threshold: 0.004
remove_borders: 4
line_threshold: 0.65
line_length_threshold: 50

point_matcher:
matcher: 0 # 0 for lightglue, 1 for superglue
image_width: 752
image_height: 480
onnx_file: "superpoint_lightglue.onnx"
engine_file: "superpoint_lightglue.engine"

keyframe:
min_init_stereo_feature: 30
lost_num_match: 10
min_num_match: 30
max_num_match: 80
tracking_point_rate: 0.65
tracking_parallax_rate: 0.1

optimization:
tracking:
mono_point: 50
stereo_point: 75
mono_line: 50
stereo_line: 75
rate: 0.5
backend:
mono_point: 50
stereo_point: 75
mono_line: 50
stereo_line: 75
rate: 0.5

ros_publisher:
feature: 1
feature_topic: "/AirSLAM/feature"
frame_pose: 1
frame_pose_topic: "/AirSLAM/frame_pose"
frame_odometry_topic: "/AirSLAM/LatestOdometry"
keyframe: 1
keyframe_topic: "/AirSLAM/keyframe"
path_topic: "/AirSLAM/odometry"
map: 1
map_topic: "/AirSLAM/map"
mapline: 1
mapline_topic: "/AirSLAM/mapline"
reloc: 0
reloc_topic: "/AirSLAM/reloc"

实际运行的结果是这样的还会出现这样的报错，而且keyframe NUM很少。