Highlight Clark comments

Results 22 comments of


                                            Highlight Clark

when testing with IMU no certain files

Change the path to match your own machine, please. e.g. ``` fgw.open("/home/YOUR_MACHINE/gw.txt"); ```

when testing with IMU no certain files

@zhm1995 These files are generated by this program when initialization is performed, so you do not need to download them.

-- Could NOT find CHOLMOD (missing: CHOLMOD_INCLUDE_DIR CHOLMOD_LIBRARIES)

Did you install cholmod library? For example, `sudo apt-get install libcholmod2.1.2`

When IMU is on, estimation is even worse than pure visual mode?

Yeah, you are right. I mean the first case as you described, where VIO fails, and system seems to propagate IMU integration as the estimator's output, and the trajectory flies...

When IMU is on, estimation is even worse than pure visual mode?

I see in source code that // ... **{ if (mbUseIMU == false) { LOG(INFO)

When IMU is on, estimation is even worse than pure visual mode?

Since there is an initialization process in the code, why haven't you called them yet? As shown in the above code of my last comment, you seem to omit this...

When IMU is on, estimation is even worse than pure visual mode?

@TouqeerAhmad See [https://github.com/jingpang/LearnVIORB/](url) for what you want. In this project, all need IMU reinitialization functions are provided at file Tracking.cc.

When IMU is on, estimation is even worse than pure visual mode?

@TouqeerAhmad Yeah, I feed my iPhone 6S image and IMU data to YGZ. Because the dataset is easy to handle( with smooth motion and enough textures), it runs well. But...

When IMU is on, estimation is even worse than pure visual mode?

Yeah, use Kalibr to acquire extrinsic matrix as well as IMU parameters please. Below is iphone6S dataset, with the same layout as EuRoC benchmark. [https://drive.google.com/open?id=0B7wDTHnOtxWUQjhvMWZiQ2c2d1U](https://drive.google.com/open?id=0B7wDTHnOtxWUQjhvMWZiQ2c2d1U)

When IMU is on, estimation is even worse than pure visual mode?

1. I made no change to source code, and only replace necessary parameters in config file. 2. The above trajectory I posted is from the dataset I shared and is...