Thank yo so much to update your code. This code works good except the result of gyroscope_random_walk is NaN. May I ask you the reason of it or any hint? I used xsens mti-100 to collect data for 2 hours. Here is my result. Thank you so much.

Hi:
in kalibr wiki, gyroscope_random_walk is regarded as in run-bias- instability（the lowest point in the Allan standard deviation）, but here is regarded as "fit a 1/2 line to the right side of the allan plot, Random walk is at tau=3 ".
I am confuse about that

Hello,

Thank you very much for making this tool available to public.

I am trying to better understand the results generated by this tool, and noticed in XSENS MTI-G-710 acceleration graph the line with -1/2 slope is fitting to Allan variance graph pretty well, but the line with the slope +1/2 is not fitting at all. Why does this tool insist on fitting a +1/2 line to the graph while the empirical data just doesn't look like a +1/2 line? Is the calculated random walk in this graph really useful?

Thank you very much,
Ahmad

Hi,

I have some trouble understanding the exact format of IMU's outputs, which will be the inputs of your functions. The IMU I have outputs instantaneous accelerations along three axes (the unit is m/s^2) and the angle change during a measurement cycle in three axes (the unit is m deg). I have read several materials which told me how to use Allan deviation to analyze IMU noise model, but their IMU seems to output instantaneous angle velocity (the unit is m deg/s) instead of the angle change during a cycle[1]. Would you tell me what is the exact format of IMU's outputs your scripts will accept? Are the outputs instantaneous measurements or the change during a measurement cycle?

PS: I tried to provide the inputs for you scripts with a csv file, but the figure seemed to be a little different from your example. The command window outputs several repeated messages and the figure also have many repeated messages. Would you please give me an idea about this?

Thank you.

[1] http://cache.freescale.com/files/sensors/doc/app_note/AN5087.pdf

I think this line should have log10(1) which is equal to 0 instead of log10(delta):

I used bagconvert to convert a rosbag to a mat. The size of the rosbag is 1.5GB but the size of the mat is 177Bytes. And the mat is a 0*7 array. How do I solve the problem?

The average sigma value is calculated here:

Correct me if I'm wrong, but I think you can only add sigma squared values, not sigma/.

Hi,
I have doubt on capturing the raw imu values while IMU remaining still. That means, we have to keep the IMU z-axis aligned with gravity direction. is this interpretation is correct or we can capture the values at any orientation of IMU with gravity direction.

If we can capture at any orientation, the projection of acceleration due to gravity on xy-plane will be coming as a bias in the measurements. is this interpretation is correct?

Hey,
I wanted to ask if this package is available for ROS1 Noetic?
Which is the default distribution this package runs on?

Hi there, which MATLAB script am I supposed to run after converting the Rosbag to a .mat file?

I used this program just to caluate allan var spend 30 mins, but cant work out ,i dont why need so long time

When I download the data form google of imu_mtig700.mat,
and run matlab SCRIPT_process_results.m,
I get a error :

The variable "results_ax" or class "results_ax.tau1" is not defined.
SCRIPT_process_results (line 30)
[fh1,sigma_a,sigma_ba] = gen_chart(results_ax.tau1,results_ax.sig2,...

How can I do about this error show? Thank you !

hello, I appreciate your nice work.
I am a newbie to ROS. I download your android-dataset-recorder and record images and IMU data.
I wonder how can I transfer the IMU data to a rosbag? I know there are some tools to create rosbag with both IMU data and images. But I don not know how to create a rosbag with only IMU. Could you please give me some advice?

hello sir,
i wanna try kalivr_allan with my own IMU sensor.
and i curious about gravity.

my sensor output include gavity value. like the example below.

acc_x : -0.001
acc_y : -0.003
acc_z : -0.984

gyr_x : 0.004
gyr_y : -0.001
gyr_z : -0.004

is it okay to have gravity value in my acceleration?
if it's not okay, i'm gonna fix the value.

and thanks for your wonderful work.

regards.

I have some trouble about how to interpret these graph. Can some one explain me what is this graph mean. what is cross red and blue line? or suggest me some keyword to search about this.

when I run catkin_make at catkin workspace, error happened and error message says "can't find MATLAB_ENG_LIBRARY (ADVANCED)"and other several MATLAB-libraries.

CMake Error: The following variables are used in this project, but they are set to NOTFOUND.
Please set them or make sure they are set and tested correctly in the CMake files:
MATLAB_ENG_LIBRARY (ADVANCED)
linked by target "bagconvert" in directory /home/duyl/catkin_ws/src/kalibr_allan/bagconvert
linked by target "bagconvert" in directory /home/duyl/catkin_ws/src/kalibr_allan/bagconvert
MATLAB_INCLUDE_DIR (ADVANCED)
used as include directory in directory /home/duyl/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/duyl/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/duyl/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/duyl/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/duyl/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/duyl/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/duyl/catkin_ws/src/kalibr_allan/bagconvert
MATLAB_MAT_LIBRARY (ADVANCED)
linked by target "bagconvert" in directory /home/duyl/catkin_ws/src/kalibr_allan/bagconvert
MATLAB_MEX_LIBRARY (ADVANCED)
linked by target "bagconvert" in directory /home/duyl/catkin_ws/src/kalibr_allan/bagconvert
linked by target "bagconvert" in directory /home/duyl/catkin_ws/src/kalibr_allan/bagconvert
MATLAB_MX_LIBRARY (ADVANCED)
linked by target "bagconvert" in directory /home/duyl/catkin_ws/src/kalibr_allan/bagconvert
linked by target "bagconvert" in directory /home/duyl/catkin_ws/src/kalibr_allan/bagconvert
MATLAB_UT_LIBRARY
linked by target "bagconvert" in directory /home/duyl/catkin_ws/src/kalibr_allan/bagconvert

I install matlab2014A to /home/MY_NAME/Downloads/matlab, cause there is no enough free space at/usr/local, does that matter?
Or, those libs were not installed like that?
if so, it's nice of you to show me some way to install them : )

-- MATLAB executable found: /usr/local/bin/matlab
-- MATLAB not found. Set MATLAB_ROOT
-- Will look for MATLAB libraries in
-- MATLAB not found. If you do have it, set MATLAB_ROOT and reconfigure (missing: MATLAB_ROOT MATLAB_INCLUDE_DIR MATLAB_LIBRARIES MATLAB_MEXEXT_PATH MATLAB_MEX_EXT)
-- Using these message generators: gencpp;geneus;genlisp;gennodejs;genpy
CMake Warning at /opt/ros/kinetic/share/catkin/cmake/catkin_package.cmake:166 (message):
catkin_package() DEPENDS on 'Eigen3' but neither 'Eigen3_INCLUDE_DIRS' nor
'Eigen3_LIBRARIES' is defined.
Call Stack (most recent call first):
/opt/ros/kinetic/share/catkin/cmake/catkin_package.cmake:102 (_catkin_package)
bagconvert/CMakeLists.txt:20 (catkin_package)

Hi,
I keep getting a segfault with:
Error in `../bagconvert/bagconvert': free(): invalid pointer: 0x0000000000e911a8
Any ideas what is causing this? Its not a show stopper but super annoying.

Thanks for your allan tool. I have two quetion

1.Is the IMU data format (timestamp(s not ns), acc_x(m/s^2), acc_y(m/s^2), acc_z(m/s^2) , gyro_x(rad/s), gyro_y(rad/s), gyro_z(rad/s) for allan tool ?
2.The tool will fit a -1/2(or 1/2) line to the left(or right) side of the allan plot, however the line is not tangent line. Is it right?
Thank you again.

I want to run the bagconverter on ubuntu20.04 but meet some problems.
/usr/bin/ld: CMakeFiles/bagconvert.dir/src/main.cpp.o: in function boost::filesystem::path::parent_path() const': /usr/local/include/boost/filesystem/path.hpp:808: undefined reference to boost::filesystem::path::find_parent_path_size() const'
/usr/bin/ld: CMakeFiles/bagconvert.dir/src/main.cpp.o: in function boost::filesystem::path::stem() const': /usr/local/include/boost/filesystem/path.hpp:811: undefined reference to boost::filesystem::path::stem_v3() const'
/usr/bin/ld: /usr/local/include/boost/filesystem/path.hpp:811: undefined reference to boost::filesystem::path::stem_v3() const' /usr/bin/ld: /usr/lib/gcc/x86_64-linux-gnu/11/../../../x86_64-linux-gnu/libpython3.8.so: undefined reference to XML_SetHashSalt'
collect2: error: ld returned 1 exit status
make[2]: *** [kalibr_allan-master/bagconvert/CMakeFiles/bagconvert.dir/build.make:161: /home/liuinze/kalibr_allan_ws/devel/lib/bagconvert/bagconvert] Error 1
make[1]: *** [CMakeFiles/Makefile2:610: kalibr_allan-master/bagconvert/CMakeFiles/bagconvert.dir/all] Error 2
make: *** [Makefile:141: all] Error 2

It looks liike my boost library is not be found. But it indeed finds the boost.1.71.0. So is anyone meet with the same problem like me? Thank you for any answers.

In the Kalibr repo Wiki , they talk about continous noise density, and discrete noise density. Using the graph-based method from this repo - does one get the continuous or the discrete noise density? I believe it is the continuous one, but I want to be sure.

When I start the "SCRIPT_allan_matparallel.m" script I get:
"opening the mat file.
loading timeseries.
imu frequency of 100.00.
sample period of 0.01000.
calculating allan deviation.
Error using parallel.Job/fetchOutputs (line 1286)
An error occurred during execution of Task with ID 1.

Error in SCRIPT_allan_matparallel (line 79)
r1 = fetchOutputs(j1);

Error in run (line 91)
evalin('caller', strcat(script, ';'));

Caused by:
Error using allan (line 93)
Array indices must be positive integers or logical values."

Hello, I met a error which is very confusing. I have already compiled the bagconvert. Then I ran the code below
rosrun bagconvert bagconvert imu.bag /imu
and the output in the terminal is under
[ INFO] [1571713279.117130547]: Starting up
段错误 (核心已转储)
so I use clion to debug, I find the program breaks at the line below
ros::init(argc, argv, "bagconvert")
It is so confusing. Any Advice? Thank you very much!

I used kalibr_bagcreater to create my imu.bag, and will I mistake the order of gyro data and acce data?

When I use catkin_make, it just like this-
(I have checked Findmatlab.cmake for :
find_program(MATLAB_EXE_PATH matlab
PATHS /usr/local/MATLAB/R2018a/bin
it still wrong!!!!)

• +++ processing catkin package: 'bagconvert'
  -- ==> add_subdirectory(kalibr_allan/bagconvert)
  -- Boost version: 1.58.0
  -- Found the following Boost libraries:
  -- system
  -- filesystem
  -- thread
  -- date_time
  -- chrono
  -- atomic
  -- MATLAB executable found: /usr/local/MATLAB/R2018b/bin/matlab
  -- MATLAB not found. Set MATLAB_ROOT
  -- Will look for MATLAB libraries in
  -- MATLAB not found. If you do have it, set MATLAB_ROOT and reconfigure (missing: MATLAB_ROOT MATLAB_INCLUDE_DIR MATLAB_LIBRARIES MATLAB_MEX_PATH MATLAB_MEXEXT_PATH MATLAB_MEX_EXT)
  -- Using these message generators: gencpp;geneus;genlisp;gennodejs;genpy
  CMake Warning at /opt/ros/kinetic/share/catkin/cmake/catkin_package.cmake:166 (message):
  catkin_package() DEPENDS on 'Eigen3' but neither 'Eigen3_INCLUDE_DIRS' nor
  'Eigen3_LIBRARIES' is defined.
  Call Stack (most recent call first):
  /opt/ros/kinetic/share/catkin/cmake/catkin_package.cmake:102 (_catkin_package)
  kalibr_allan/bagconvert/CMakeLists.txt:20 (catkin_package)

-- +++ processing catkin package: 'vio_data_simulation'
-- ==> add_subdirectory(vio_data_simulation-ros)
CMake Error: The following variables are used in this project, but they are set to NOTFOUND.
Please set them or make sure they are set and tested correctly in the CMake files:
MATLAB_ENG_LIBRARY (ADVANCED)
linked by target "bagconvert" in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
linked by target "bagconvert" in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
MATLAB_INCLUDE_DIR (ADVANCED)
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
used as include directory in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
MATLAB_MAT_LIBRARY (ADVANCED)
linked by target "bagconvert" in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
MATLAB_MEX_LIBRARY (ADVANCED)
linked by target "bagconvert" in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
linked by target "bagconvert" in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
MATLAB_MX_LIBRARY (ADVANCED)
linked by target "bagconvert" in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
linked by target "bagconvert" in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert
MATLAB_UT_LIBRARY
linked by target "bagconvert" in directory /home/xuhang/catkin_ws/src/kalibr_allan/bagconvert

-- Configuring incomplete, errors occurred!
See also "/home/xuhang/catkin_ws/build/CMakeFiles/CMakeOutput.log".
See also "/home/xuhang/catkin_ws/build/CMakeFiles/CMakeError.log".
Makefile:990: recipe for target 'cmake_check_build_system' failed
make: *** [cmake_check_build_system] Error 1

The plots should be of the Allan deviation, rather than the variance, so a square root must be taken of the variance for plotting. However, this is done twice in this tool: once just before the end of allan.m, and once in the plot function (loglog()). Thus, the final graph is incorrect, because what is plotted is the square root of the Allan deviation, rather than the Allan deviation.

We know many sensors have 3 axises accelemeters and 3 axised gyros. I see the results in the figure only have a set of random walk and white noise. Should we consider the other two aixes' noise parameter?

I mean we cannot assume all three axises have the same noise parameter. Isn't it? :(

Hi,

Is there any mathematical approach to find out how many hours of data should be recorded depending on the application? For example if this calibration is going to be used on a drone that flies only for 1 minute, once a day, do we still need to record and process 3-4 hours of data?

Thank You,

I use the xsens-MTI-100 dataset download from here.The result of accelerometer_random_walk is NaN.

Here is the terminal output.

>> SCRIPT_process_results
=> opening the mat file.
=> plotting accelerometer.
tau = 1.00 | tauid1 = 1190
h_fit1 slope = -0.5000 | y-intercept = -6.7606
h_fit2 slope = 0.5000 | y-intercept = NaN
tau = 2.99 | tauid2 = 1304
=> plotting gyroscope.
tau = 1.00 | tauid1 = 1190
h_fit1 slope = -0.5000 | y-intercept = -8.7292
h_fit2 slope = 0.5000 | y-intercept = -12.9155
tau = 2.99 | tauid2 = 1304
=> final results
accelerometer_noise_density = 0.00115858
accelerometer_random_walk   = NaN
gyroscope_noise_density     = 0.00016179
gyroscope_random_walk       = 0.00000426

I am not familiar with the Allan variance. Could you give me some suggestions?
Thanks a lot!