That looks awesome. Any idea why tango failed in this run? Seems it's doing pretty well except when the door is being opened, which introduces outliers in tracking. But msckf should have the potential to implicitly reject them. Btw, what's your underlying filter, msckf or ekf?
Awesome! Good job!
Awesome.
That looks awesome. Any idea why tango failed in this run? Seems it's doing pretty well except when the door is being opened, which introduces outliers in tracking. But msckf should have the potential to implicitly reject them. Btw, what's your underlying filter, msckf or ekf?
My guess is that it doesn't have loop closure (doesn't adjust past position estimates based on recognizing past scenes again).
What's up Xiaohan! :D
What did you use as visual inertial sensor?
VINS with IMU OR NOT?
VINS: IN is short of inertial, which uses IMU measurement.
I can't run on Android studio. Anyone can help?
Any Android porting?