vslam_node

A collection of wrappers (ROS nodes) and tools to test visual / visual-inertial SLAM algorithms

The code that can prepare the datasets that required by the following packages can be found in https://github.com/rising-turtle/img_pub.

ORB_VIO

The original VIORB package is https://github.com/jingpang/LearnVIORB
To enable the system starts from the first frame, use depth to initialize features just for the first frame, modification are made in https://github.com/rising-turtle/LearnVIORB/tree/master
off_viorb.cpp read rgb-d and imu data from disk.
off_viorb_dpt_first.cpp use the depth data for the first frame.

VINS-Mono

The original VINS-Mono package is https://github.com/HKUST-Aerial-Robotics/VINS-Mono
Add launch and config file to run VINS-Mono using our dataset.
Also, vins-mapping.cpp use the result of VINS-Mono and the camera's data (rgb-d) to generate trajectory.ply and map.ply.

DSO

The original dso_ros package is https://github.com/JakobEngel/dso_ros
The original dso package is https://github.com/JakobEngel/dso
compile it in ROS, plus add config and launch file to run it using our dataset.
main.cpp dso_live
main_dso_pangolin.cpp dso_offline

graph

A simple gtsam based visual-inertial SLAM implementation, following the paper

• Information fusion in navigation systems via factor graph based incremental smoothing, V. Indelman, S. Williams, M. Kaess, and F. Dellaert, Robotics and Autonomous Systems, vol. 61, no. 8, pp. 721–738, 2013

imu_img_show

A simple tool to publish the angular change (delta_roll,pitch,yaw) from imu's gyro and meanwhile publish camera's image.
Another package can receive the published data and, show the gyro's rotation and meanwhile display the camera data. https://github.com/rising-turtle/gyro_display
In this way, we can visually see whether and how the data from IMU and camera have been synchronized.

imu_pub

A simple IMU reader and pulisher. And a simple test to compute the gravity direction using the first several IMU readings.

msf-multi sensor fusion

The original MSF package is https://github.com/ethz-asl/ethzasl_msf
Add some launch files to use SVO as vision-only part in MSF.

okvis

The original okvis package is https://github.com/ethz-asl/okvis
Add source, config, launch files to process the R200 dataset

svo

The original svo package is https://github.com/uzh-rpg/rpg_svo
Modified version enables to add prior pose, is https://github.com/rising-turtle/rpg_svo
Add source files:
vo_node.cpp: a ros node for svo
vo_node_dpt.cpp: use depth of the first frame to solve scale, based on vo_node.cpp
test_pipeline_visual.cpp: run svo using its own dataset: sin2_tex2_h1_v8_d
feature_detect.cpp: extract and show fast features
good_feature_for_track.cpp: extract and show 'good' features, which are the same features used in VINS-Mono
vio_node_dpt.cpp: graph based loosely-coupled vio, use svo as the vision-only part
msckf_node.cpp: msf based loosely-coupled vio, use svo as the vision-only part
msckf_node_dpt.cpp: use depth of the first frame to solve scale, based on msckf_node.cpp