Skip to content

use_imu.md

Latest commit

 

History

History
66 lines (43 loc) · 3.05 KB

use_imu.md

File metadata and controls

66 lines (43 loc) · 3.05 KB

How to run the system with imu data

EuRoC dataset

# Tracker (choose one from the following two cases)
# Case1: Monocular input with IMU data
ros2 launch univloc_tracker euroc_mono.launch.py camera_setup:=Monocular_Inertial publish_tf:=false clean_keyframe:=true

# Case2: Stereo input with IMU data
ros2 launch univloc_tracker euroc_stereo.launch.py camera_setup:=Stereo_Inertial publish_tf:=false clean_keyframe:=true

# Server
ros2 launch univloc_server server.launch.py fix_scale:=true correct_loop:=false

# Play rosbag
ros2 bag play xxx

Explanations for the flags

  • clean_keyframe: whether to clean redundant keyframes in the map database. The performance may vary between sequences in EuRoC dataset when setting clean_keyframe as true or false.
  • fix_scale: whether to fix scale during server optimization. If imu data is used, then fix_scale is always true since absolute scale can be obtained either by imu (Monocular_Inertial), or by depth or stereo camera (RGBD_Inertial, Stereo_Inertial).
  • correct_loop: whether to enable loop correction in server. The performance may vary between sequences in EuRoC dataset when setting correct_loop as true or false.

OpenLORIS dataset (RGBD input)

# Tracker
# Please change the value of camera_fps that represents the real FPS of input camera images, the default value is 30.0 if you don't specify it
ros2 launch univloc_tracker tracker.launch.py camera:=d400 publish_tf:=false queue_size:=0 camera_setup:=RGBD_Inertial get_camera_extrin_from_tf:=false accel_topic:=/d400/accel/sample gyro_topic:=/d400/gyro/sample camera_fps:=30.0

# Server
ros2 launch univloc_server server.launch.py fix_scale:=true correct_loop:=false

# Play rosbag
ros2 bag play xxx

Explanations for the flag

  • get_camera_extrin_from_tf: whether to get camera extrinsic matrix from tf. Since OpenLORIS dataset cannot obtain extrinsic matrix, set to false to avoid unnecessary waiting.

Self-recorded Dataset

You need to provide the following imu-related parameters to the system:

  1. Transform between camera and imu: get_imu_extrin_from_tf is used to control whether to obtain from tf or enter the values directly in tracker.yaml. When set to true, you must provide the correct imu_frame and image_frame from tf tree; When set to false, put the values into tf_camera_imu.

  2. Feeding imu data (two ways are supported): 1) Providing the topic of imu_topic if accel and gyro data are already combined into one single topic; 2) Providing both the topics of accel_topic and gyro_topic if accel and gyro data are separated topics.

  3. Parameters of IMU sensor: the below three params should be obtained from the manufacturer of sensor. Note that tracker.yaml provides default values for RealSense D435i camera. If you are not able to obtain these values, you can use the default ones.

  • imu_frequency
  • imu_noise
  • imu_bias_walker

Limitations

  • RGBD_Inertial mode is not extensively tested yet, only verified on OpenLORIS market dataset.