I try to calibrate a MPU6050 with a depth camera; I got the good state for the camera calibration alone, with reprojection erros to be about 0.4-0.5 pixel ( in the moving state);
But when I try to calibrate the camera and IMU as a whole, the camera reprojection error is too big ( to nearly 3-4 pixels). I try to increase the noise property of the IMU (with 2x noise density and 10x randwalk), but the result does not get any improvent.
I am a little confused by the following optimization options. Could someone share some light on it ,so that I can get better simulation results between body-estimation and measured data?
iCal.buildProblem(splineOrder=6,
poseKnotsPerSecond=100,
biasKnotsPerSecond=100,
doPoseMotionError=False,
doBiasMotionError=True,
blakeZisserCam=-1,
huberAccel=-1,
huberGyro=-1,
noTimeCalibration=parsed.no_time,
noChainExtrinsics=(not parsed.recompute_chain_extrinsics),
maxIterations=parsed.max_iter,
timeOffsetPadding=parsed.timeoffset_padding,
verbose = parsed.verbose)
Hi friends:
I try to calibrate a MPU6050 with a depth camera; I got the good state for the camera calibration alone, with reprojection erros to be about 0.4-0.5 pixel ( in the moving state); But when I try to calibrate the camera and IMU as a whole, the camera reprojection error is too big ( to nearly 3-4 pixels). I try to increase the noise property of the IMU (with 2x noise density and 10x randwalk), but the result does not get any improvent.
Here is the calibration result. Any suggestions for the further modification? Sincere thanks in advance! test_420_kalibr-report-cam.pdf test_420_kalibr-report-imucam.pdf