Observability-Aware Self-Calibration of Visual and Inertial Sensors for Ego-Motion Estimation

Abstract

External effects such as shocks and temperature variations affect the calibration of visual-inertial sensor systems and thus they cannot fully rely on factory calibrations. Recalibrations performed on short user-collected datasets might yield poor performance since the observability of certain parameters is highly dependent on the motion. Additionally, on resourceconstrained systems (e.g mobile phones), full-batch approaches over longer sessions quickly become prohibitively expensive.In this paper, we approach the self-calibration problem by introducing information theoretic metrics to assess the information content of trajectory segments, thus allowing to select the most informative parts from a dataset for calibration purposes. With this approach, we are able to build compact calibration datasets either: (a) by selecting segments from a long session with limited exciting motion or (b) from multiple short sessions where a single sessions does not necessarily excite all modes sufficiently. Realworld experiments in four different environments show that the proposed method achieves comparable performance to a batch calibration approach, yet, at a constant computational complexity which is independent of the duration of the session.

Visual-Inertial Navigation: A Concise Review

“…Qin and Shen [132] extended their prior work on batch-based monocular VINS [31] to include the time offset between the camera and IMU by interpolating the locations of features on the image plane. Schneider et al [140] proposed the observability-aware online calibration utilizing the most informative motions. While we recently have also analyzed the degenerate motions of spatiotemporal calibration [141], it is not fully understood how to optimally model intrinsics and simultaneously calibrate them along with extrinsics [142,143].…”

Optimization-Based Online Initialization and Calibration of Monocular Visual-Inertial Odometry Considering Spatial-Temporal Constraints

“…In their extended batch-based monocular VINS (termed VINS-Mono) [40], the IMU biases are included in a sliding window nonlinear estimator. Schneider et al [41] introduced information-theoretic metrics to assess the information content of trajectory segments, thus allowing them to select the most informative parts from a dataset for extrinsic spatial calibration purposes. Huang et al [19] proposed an estimator to incrementally solve several linear equations to estimate the spatial parameters between an IMU and two cameras.…”

Extrinsic Visual–Inertial Calibration for Motion Distortion Correction of Underwater 3D Scans

“…In [43], the trajectory is parameterized with B-splines and included in the optimization problem. In [44], only the parts of the trajectory that contribute more to the observability of the problem are considered, so that the computational complexity is reduced. Other interesting works can be found in [45][46][47].…”