We propose a continuous-time spline-based formulation for visual-inertial odometry (VIO). Specifically, we model the poses as a cubic spline, whose temporal derivatives are used to synthesize linear acceleration and angular velocity, which are compared to the measurements from the inertial measurement unit (IMU) for optimal state estimation. The spline boundary conditions create constraints between the camera and the IMU, with which we formulate VIO as a constrained nonlinear optimization problem. Continuous-time pose representation makes it possible to address many VIO challenges, e.g., rolling shutter distortion and sensors that may lack synchronization. We conduct experiments on two publicly available datasets that demonstrate the state-of-the-art accuracy and real-time computational efficiency of our method.
Velocity Continunity Constraint
Rotation Continunity Constraint
The proposed VIO system running on MH1 of EuRoC dataset. The estimated trajectory (red) is well aligned (by SE(3)) to the ground-truth poses (green). The scale is also recovered accurately.
RMSEs (in meters) of 10 runs (rows) for different methods on each sequence (columns) from EuRoC dataset (MH, V1, V2) and TUM VI dataset (TR). DSO results are aligned with Sim(3); VI-DSO and SplineVIO are aligned with SE(3).
The median RMSEs over 10 runs for different methods on each sequence from EuRoC dataset and TUM VI dataset. For VI-DSO,  are the results of the 3rd party implementation;  are the original results reported in the paper , we include these results for reference.