When is time stamp of frame captured?

[reinzler]

I have to know precision time of the end of exposure, how can i get it?
I saw get_timestamp() function but i'm not sure that it is exactly what i need.

[MartyG-RealSense]

Hi @reinzler There are a range of different timestamps that have different characteristics. At #2188 (comment) a RealSense team member provides a list of timestamp types and descriptions for them, and further information about how the timestamps work at #4525

In regard to references in these lnks about Global Time (which is enabled by default on 400 Series cameras), information can be found at #3909

Documentation for the get_timestamp() instruction can be found at the link below.

https://intelrealsense.github.io/librealsense/doxygen/classrs2_1_1frame.html#a25f71d45193f2f4d77960320276b83f1

My understanding from the above information resources is that if hardware timestamps are available then get_timestamp() provides the frame_timestamp, which designates the beginning of UVC frame transmission (the first USB chunk sent towards host) and readout begins after exposure is completed.

[reinzler]

Thanks for fast responce, how can I get FRAME_TIMESTAMP using Python?

[MartyG-RealSense]

First, disable Global Time using the Python code at #9172 (comment)

You should be able to obtain frame_timestamp using the Python code below.

frames.get_frame_metadata(rs.frame_metadata_value.frame_timestamp)

An example of Python code for printing the frame_timestamp metadata value is at #9891

[reinzler]

Thanks, how I can convert the Hardware clock to epochs to compare with the global time?

[MartyG-RealSense]

The section of the global timestamp system's code that makes use of EPOCH is at the link below.

https://github.com/IntelRealSense/librealsense/blob/master/src/global_timestamp_reader.cpp#L208-L211

My understanding from IntelRealSense/realsense-ros#1906 (comment) is that global time is not a direct conversion of the hardware time, but instead uses the system clock as a means to perform correction of the hardware timestamp.

[MartyG-RealSense]

Hi @reinzler Was the information in the above comment helpful to you, please? Thanks!

[reinzler]

Yes, it was helpful, thanks, another question - is it possible to fuse data from 2 different IMU ( VN100, Bosh BMI055), GPS (HERE 3) and SLAM from IntelRealSense D455 using ROS robot_localization package?

[MartyG-RealSense]

Multiple sensors, such as multiple IMUs, can be fused in robot_localization using state estimation nodes, as described at the robot_localization documentation link below.

http://docs.ros.org/en/kinetic/api/robot_localization/html/configuring_robot_localization.html#the-differential-and-relative-parameters

Intel's RealSense SLAM guide for ROS makes use of robot_localization, though there have not been previous cases of the SLAM data being used with sensors other than a single IMU-equipped RealSense camera.

https://github.com/IntelRealSense/realsense-ros/wiki/SLAM-with-D435i

[reinzler]

Thank you very much for fast and useful response!

[reinzler]

Thank you for your help, a have another question, can I pass some prepared in advance data from D455 to ROS robot_localization library to produce some simulation?

[MartyG-RealSense]

Intel's SLAM guide suggests replaying a saved RealSense bag file by setting the rosparam use_sim_time to true and then reading the bag with rosbag play

roscore >/dev/null 2>&1 &
rosparam set use_sim_time true
rosbag play my_bagfile_1.bag --clock
roslaunch realsense2_camera opensource_tracking.launch offline:=true

The link below suggests a method for loading a bag file directly into robot_localization.

https://answers.ros.org/question/268046/robot_localization-and-bag-file-output/?answer=268228#post-id-268228

[MartyG-RealSense]

Hi @reinzler Do you require further assistance with this case, please? Thanks!

[MartyG-RealSense]

Case closed due to no further comments received.