MIT-SPARK
IMU Integration
I have a problem while integrating my sensor yaml files with Kimera-VIO as shown in the below text
No vio parameters specified, using default. Applying IMU timestamp shift of: 3.59449e+246ns. &&&&&&&&&&&&&&&&&&&& TRACKER PARAMETERS &&&&&&&&&&&&&&&&&&&&&&&&&&& ** Feature tracking parameters ** klt_win_size_: 24 klt_max_iter_: 30 klt_max_level_: 4 klt_eps_: 0.1 maxFeatureAge_: 15 ** Feature detection parameters ** maxFeaturesPerFrame_: 800 quality_level_: 0.001 min_distance_: 20 block_size_: 3 use_harris_detector_: 0 k_: 0.04 ** Sparse Stereo Matching parameters ** equalize_image_: 0 nominalBaseline_: 0.11 vision_sensor_type_: 0 toleranceTemplateMatching_: 0.15 templ_cols_: 101 templ_rows_: 11 stripe_extra_rows_: 0 minPointDist_: 0.5 maxPointDist_: 10 bidirectionalMatching_: 0 subpixelRefinementStereo_: 0 ** Feature selection parameters ** featureSelectionCriterion_: 0 featureSelectionHorizon_: 3 featureSelectionNrCornersToSelect_: 600 featureSelectionImuRate_: 0.005 featureSelectionDefaultDepth_: 2 featureSelectionCosineNeighborhood_: 0.984808 featureSelectionUseLazyEvaluation_: 1 useSuccessProbabilities_: 1 ** RANSAC parameters ** useRANSAC_: 1 minNrMonoInliers_: 10 minNrStereoInliers_: 5 ransac_threshold_mono_: 1e-06 ransac_threshold_stereo_: 1 ransac_use_1point_stereo_: 1 ransac_use_2point_mono_: 1 ransac_max_iterations_: 100 ransac_probability_: 0.995 ransac_randomize_: 0 ** STEREO tracker parameters ** intra_keyframe_time_: 0.2 minNumberFeatures_: 0 useStereoTracking_: 1 ** OTHER parameters ** disparityThreshold_: 0.5 &&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&& LoopClosureDetector:: Loading vocabulary from ../vocabulary/ORBvoc.yml Loaded vocabulary with 971815 visual words. Spinning Visualizer. Shutting down VIO pipeline once processing has finished. Frontend launched (parallel_run set to 1). ------------------- Initialize Pipeline with frame k = 50-------------------- Spinning StereoFrontEnd. getRectifiedImages: abnormal baseline: 0, nominalBaseline: 0.11(+/-10%)
Using Regular VIO with modality 0 [0, 0, 0]';isam_param type: ISAM2GaussNewtonParams wildfireThreshold: -1 relinearizeThreshold: 0.01 relinearizeSkip: 1 enableRelinearization: 1 evaluateNonlinearError: 0 factorization: CHOLESKY cacheLinearizedFactors: 1 enableDetailedResults: 0 enablePartialRelinearizationCheck: 0 findUnusedFactorSlots: 1 Initial state seed:
- Initial pose: |0.005, 5.4873e+247, 2.51834e-316| |0.133927, 6.52016e+252, 2.56383e-316| |1.61701, 1.60426e-316, 1.00444e-320|
[-nan, 0, 0]';
- Initial vel: 4.94066e-324 000000000000 000000000000
- Initial IMU bias: acc = [0, 0, 0]' gyro = [0, 0, 0]'
Indeterminant linear system detected while working near variable 8646911284551352320 (Symbol: x0).
Thrown when a linear system is ill-posed. The most common cause for this error is having underconstrained variables. Mathematically, the system is underdetermined. See the GTSAM Doxygen documentation at http://borg.cc.gatech.edu/ on gtsam::IndeterminantLinearSystemException for more information. ERROR: Variable has type 'x' and index 0 Smoother's factors: [ size: 3
Factor 0: PriorFactor on x0 prior mean: R: [ 0.005 5.48729961e+247 2.51834192e-316; 0.13392721 6.52016458e+252 2.56383233e-316; 1.6170107 1.6042604e-316 1.00443546e-320 ] [-nan, 0, 0]'; noise model: Gaussian[ -nan -nan -nan -nan -nan -nan; 0 -nan -nan -nan -nan -nan; 0 0 -nan -nan -nan -nan; 0 0 0 -nan -nan -nan; 0 0 0 0 -nan -nan; 0 0 0 0 0 -nan ]
Factor 1: PriorFactor on v0 prior mean: [ 4.94065646e-324; 0; 0 ] isotropic dim=3 sigma=0.001
Factor 2: PriorFactor on b0 prior mean: acc = [0, 0, 0]' gyro = [0, 0, 0]' noise model: diagonal sigmas[0.1; 0.1; 0.1; 0.01; 0.01; 0.01];
] State values [ Values with 0 values: ] =============== START:CATCHING EXCEPTION =============== Nr of factors in graph (smoother getFactors): 3, with factors: [
] Nr of new factors to add: 0 with factors: [ (slot # wrt to new_factors_tmp graph)
] Nr deleted slots: 0, with slots: [
] Nr of values in state_ : 0, with keys: [
] Nr values in new_values_ : 0, with keys: [
] =============== END: CATCHING EXCEPTION ===============
Uploaded as .txt files but the content is the same as my yaml files with the same file names provided in the repo
sensor_imu.txt sensor_cam0.txt](https://github.com/MIT-SPARK/Kimera-VIO/files/3874334/sensor_cam0.txt)
@omarosamahu we have changed extensively the API for Kimera-VIO so the above does not necessarily apply anymore.
In particular, now we parse our own ImuParams.yaml file inside the params folder (so to avoid problems like this one).
One thing that scaped my sight first is that it seems that there is an IMU timeshift being applied (which is only there in case IMU and camera clocks are not perfectly synchronized), but in your param file it seems as if the shift wasn't initialized to 0: I'm referring to this line Applying IMU timestamp shift of: 3.59449e+246ns. (the second one in the first message of this thread).