Convergence of trajectory

[msh003]

Hello,

Thank you for sharing this impressive work on GitHub. I’ve been testing the package but encountered an issue with convergence in the estimated trajectory. Specifically, I tested it using the ''HKU_HDR_circle.bag'' dataset with both the normal version and the flight version.

In the normal version, I encountered the following message:

[ INFO] [1732747041.557555366]:  big IMU acc bias estimation 3.813893
[ INFO] [1732747041.636855711]:  big IMU acc bias estimation 3.820456
[ INFO] [1732747041.699925838]:  big IMU acc bias estimation 3.832804
[ INFO] [1732747041.761157015]:  big IMU acc bias estimation 3.828428
[ INFO] [1732747041.831440810]:  big IMU acc bias estimation 3.784428
[ INFO] [1732747041.906270996]:  big IMU acc bias estimation 3.788193
[ INFO] [1732747041.975876834]:  big IMU acc bias estimation 3.787715

and this is the output:

For the flight version, I face this message:

[ WARN] [1732747322.220953571]:  little feature 0
[ WARN] [1732747322.279084792]:  little feature 0
[ WARN] [1732747322.339658996]:  little feature 0
[ WARN] [1732747322.417510261]:  little feature 0
[ WARN] [1732747322.467794343]:  little feature 0
[ WARN] [1732747322.545610585]:  little feature 0
[ WARN] [1732747322.620690236]:  little feature 0
[ WARN] [1732747322.670175536]:  little feature 0
[ WARN] [1732747322.741725609]:  little feature 0
[ WARN] [1732747322.828514450]:  little feature 0
[ WARN] [1732747322.874121164]:  little feature 0
[ WARN] [1732747322.937509904]:  little feature 0
[ WARN] [1732747323.026348289]:  little feature 0
[ WARN] [1732747323.071452277]:  little feature 0
[ WARN] [1732747323.139134734]:  little feature 0
[ WARN] [1732747323.213702511]:  little feature 0
[ WARN] [1732747323.270878466]:  little feature 0

and this the output:

I would appreciate it if you please help me how to fix this issue.

[cpymaple]

Hi, sorry for the late reply. For the hku stereo dataset, you should use the flight_version. Is the flight_version config work successfully on other hku stereo dataset(such as hku aggressive translation)?

Hello,

Thank you for sharing this impressive work on GitHub. I’ve been testing the package but encountered an issue with convergence in the estimated trajectory. Specifically, I tested it using the ''HKU_HDR_circle.bag'' dataset with both the normal version and the flight version.

In the normal version, I encountered the following message:

[ INFO] [1732747041.557555366]:  big IMU acc bias estimation 3.813893
[ INFO] [1732747041.636855711]:  big IMU acc bias estimation 3.820456
[ INFO] [1732747041.699925838]:  big IMU acc bias estimation 3.832804
[ INFO] [1732747041.761157015]:  big IMU acc bias estimation 3.828428
[ INFO] [1732747041.831440810]:  big IMU acc bias estimation 3.784428
[ INFO] [1732747041.906270996]:  big IMU acc bias estimation 3.788193
[ INFO] [1732747041.975876834]:  big IMU acc bias estimation 3.787715

and this is the output:

For the flight version, I face this message:

[ WARN] [1732747322.220953571]:  little feature 0
[ WARN] [1732747322.279084792]:  little feature 0
[ WARN] [1732747322.339658996]:  little feature 0
[ WARN] [1732747322.417510261]:  little feature 0
[ WARN] [1732747322.467794343]:  little feature 0
[ WARN] [1732747322.545610585]:  little feature 0
[ WARN] [1732747322.620690236]:  little feature 0
[ WARN] [1732747322.670175536]:  little feature 0
[ WARN] [1732747322.741725609]:  little feature 0
[ WARN] [1732747322.828514450]:  little feature 0
[ WARN] [1732747322.874121164]:  little feature 0
[ WARN] [1732747322.937509904]:  little feature 0
[ WARN] [1732747323.026348289]:  little feature 0
[ WARN] [1732747323.071452277]:  little feature 0
[ WARN] [1732747323.139134734]:  little feature 0
[ WARN] [1732747323.213702511]:  little feature 0
[ WARN] [1732747323.270878466]:  little feature 0

and this the output:

I would appreciate it if you please help me how to fix this issue.

[msh003]

Dear Dr.Chen,
Thank you for your attention, and I apologize for the delay in my response. I tested the ESIO algorithm using the HKU_aggressive_translation.bag file. To configure the "Flight Version," I made the following changes in the stereo_estimator_node.cpp file:

Commented out the following lines (normal version):
Lines 56–59
Lines 68–71
Lines 303–308
Lines 333–338

Uncommented the following lines (flight version):
Lines 62–64
Lines 74–76
Lines 311–316
Lines 341–346.

There is huge difference between the estimated trajectory and ground truth, as you can see in this figure.

Also, I see these lines in the terminal during running the package
[ INFO] [1733332959.745664285]: big IMU acc bias estimation 2.580365 [ INFO] [1733332959.845557111]: big IMU acc bias estimation 2.552867 [ INFO] [1733332959.887620998]: big IMU acc bias estimation 2.568387 [ INFO] [1733332959.944373545]: big IMU acc bias estimation 2.530736 [ WARN] [1733332962.391862539]: little feature 0 [ WARN] [1733332962.450988316]: little feature 0 [ WARN] [1733332962.510605813]: little feature 1 [ WARN] [1733332962.607228985]: little feature 0 [ WARN] [1733332962.646818703]: little feature 1 [ WARN] [1733332962.719043507]: little feature 1
I am using ROS Noetic on Ubuntu 20.04 and have installed Ceres 1.14.0. I would greatly appreciate your guidance in resolving this issue.

[cpymaple]

Dear Dr.Chen, Thank you for your attention, and I apologize for the delay in my response. I tested the ESIO algorithm using the HKU_aggressive_translation.bag file. To configure the "Flight Version," I made the following changes in the stereo_estimator_node.cpp file:

Commented out the following lines (normal version): Lines 56–59 Lines 68–71 Lines 303–308 Lines 333–338

Uncommented the following lines (flight version): Lines 62–64 Lines 74–76 Lines 311–316 Lines 341–346.

There is huge difference between the estimated trajectory and ground truth, as you can see in this figure.

Also, I see these lines in the terminal during running the package [ INFO] [1733332959.745664285]: big IMU acc bias estimation 2.580365 [ INFO] [1733332959.845557111]: big IMU acc bias estimation 2.552867 [ INFO] [1733332959.887620998]: big IMU acc bias estimation 2.568387 [ INFO] [1733332959.944373545]: big IMU acc bias estimation 2.530736 [ WARN] [1733332962.391862539]: little feature 0 [ WARN] [1733332962.450988316]: little feature 0 [ WARN] [1733332962.510605813]: little feature 1 [ WARN] [1733332962.607228985]: little feature 0 [ WARN] [1733332962.646818703]: little feature 1 [ WARN] [1733332962.719043507]: little feature 1 I am using ROS Noetic on Ubuntu 20.04 and have installed Ceres 1.14.0. I would greatly appreciate your guidance in resolving this issue.

Hi, for the ESIO version, you should revise the corresponding codes in "esio_estimator_node.cpp" file. (ESVIO need to revise the stereo_estimator_node.cpp file). It seems that you do not revise esio_estimator_node.cpp file.

[msh003]

Dear Dr. Chen,

Thank you for your swift response.
I revised the esio_estimator_node.cpp file by uncommenting the lines related to the flight version and commenting out the lines related to the normal version. After making these changes, I replayed the HKU_aggressive_translation.bag file. The estimation has improved compared to before, but I suspect there may still be an issue. Please find the output image attached.
I would appreciate it if you could verify whether this output matches the results you obtained with the code.

Also, I got this lines in terminal when I ran ''roslaunch esvio_estimator esio.launch '':

[ INFO] [1733595331.784810150]:  big IMU acc bias estimation 2.766016
[ INFO] [1733595331.869238103]:  big IMU acc bias estimation 2.754046
[ INFO] [1733595331.944934899]:  big IMU acc bias estimation 2.661115
[ INFO] [1733595331.989086742]:  big IMU acc bias estimation 2.659963
[ INFO] [1733595332.076539383]:  big IMU acc bias estimation 2.576063
[ INFO] [1733595332.143106628]:  big IMU acc bias estimation 2.503940
[ WARN] [1733595335.923615790]:  little feature 0
[ WARN] [1733595356.944553702]:  little feature 0
[ WARN] [1733595363.191170806]:  little feature 1
[ WARN] [1733595364.036827102]:  little feature 1
[ WARN] [1733595364.095857791]:  little feature 0
[ WARN] [1733595364.565762369]:  little feature 1
[ WARN] [1733595365.519832918]:  little feature 1
[ WARN] [1733595369.381125480]:  little feature 1

Thank you so much for your time and assistance.

[cpymaple]

Dear Dr. Chen,

Thank you for your swift response. I revised the esio_estimator_node.cpp file by uncommenting the lines related to the flight version and commenting out the lines related to the normal version. After making these changes, I replayed the HKU_aggressive_translation.bag file. The estimation has improved compared to before, but I suspect there may still be an issue. Please find the output image attached. I would appreciate it if you could verify whether this output matches the results you obtained with the code.

Also, I got this lines in terminal when I ran ''roslaunch esvio_estimator esio.launch '':

[ INFO] [1733595331.784810150]:  big IMU acc bias estimation 2.766016
[ INFO] [1733595331.869238103]:  big IMU acc bias estimation 2.754046
[ INFO] [1733595331.944934899]:  big IMU acc bias estimation 2.661115
[ INFO] [1733595331.989086742]:  big IMU acc bias estimation 2.659963
[ INFO] [1733595332.076539383]:  big IMU acc bias estimation 2.576063
[ INFO] [1733595332.143106628]:  big IMU acc bias estimation 2.503940
[ WARN] [1733595335.923615790]:  little feature 0
[ WARN] [1733595356.944553702]:  little feature 0
[ WARN] [1733595363.191170806]:  little feature 1
[ WARN] [1733595364.036827102]:  little feature 1
[ WARN] [1733595364.095857791]:  little feature 0
[ WARN] [1733595364.565762369]:  little feature 1
[ WARN] [1733595365.519832918]:  little feature 1
[ WARN] [1733595369.381125480]:  little feature 1

Thank you so much for your time and assistance.

Hi, I have already test the codes with the above suggestions, and it works normally on my NUC computer. 1. Have you change other config parameter or other codes? Maybe you can clone the codes again and only revise the "esio_estimator_node.cpp" file. 2. Which CPU of your computer? The computing power may have an effect on the results.