Problems for applying AirSim to Visual Inertial Odometry

[ShukuiZhang]

I know the following questions are a bit too much but that is all that I concerned. I will appreciate it if anyone can answer even one of them.
The questions are:

1. I need scene and depth (or disparity) image with synchronization, but it seems to be an issue for now. See issue. Even that is intractable for now, how can I get the "exact simulator time" when the images and the pose (e.g. position, orientation) are obtained?

2. How can I get the ground truth intrinsic parameters for the LeftPIP camera. See issue. The point_cloud code works only for default depth window size, but unfortunately, I need larger image than that.

3. If I understand correctly, the return value of getPosition(), getOrientation() are position and orientations for the "drone", but not for the camera. If I need to know the position and orientation of the camera, I need the relative pose between drone and camera. But seems no clue.

4. I noticed there is an IMU module, which is really great for me. However, the api does not provide a function to get the imu measurements (e.g. acceleration, angular velocity), is there any sample code that shows how to set it up and work.

My greatest appreciation!!

[wingver]

Hi Zhang, I'm learning VIO and put effort into the same thing you mentioned above. Maybe we can work together to make things done.Can you please contact me, email or wechat？

[ShukuiZhang]

Hi Ye, Nice to hear you are in this area. The problems I posted the other day are progressing! You can keep an eye on new issues because I created new threads for some of them and they get updated! Are you a student?

…

On Tue, Jul 4, 2017 at 8:54 PM, Ye Yu ***@***.***> wrote: Hi Zhang, I'm learning VIO and put effort into the same thing you mentioned above. Maybe we can work together to make things done.Can you please contact me, email or wechat？ — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#309 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AYBp9AUb63pbvPIE89k1DrzLH7_flIBzks5sKwjkgaJpZM4OJyzB> .

-- Best Regards, Shukui Zhang Ph.D. Candidate Electrical and Computer Engineering, University of California, Riverside

[sytelus]

I've added new APIs that answers 3 out of 4 questions :). The API simGetImages() allows you to get left, right and depth images simultaneously. The API return value is a struct that contains camera position, orientation and time stamp.

For intrinsic parameters of camera, see this discussion. Basically you need to convert depth to camera plan and then use FOV = 90 to create matrix. This code has example of doing many of this things.

[sytelus]

FYI – I just added ability to get camera parameters via APIs in AirSim (1st bullet point below). Here’s more info in our docs.