Observability-Enhancement Boresight Calibration of Camera-IMU System: Theory and Experiments

Airborne target tracking with vision sensors is one of the indispensable and critical components in unmanned aerial vehicles (UAVs). It is critical to calibrate the misalignment angles between the onboard camera and inertial measurement unit (IMU) to improve the accuracy of target tracking. In this article, the observability of boresight misalignment angle estimation using a cooperative target is first analyzed. Based on the observability Gramian, an optimization criterion for maximizing the degree of observability is determined, and the optimal geometries between the UAV and the target are derived. Taking the physical constraints of the UAV into consideration, an online trajectory optimization approach for a UAV with a vision sensor is proposed to achieve in-flight calibration of the misalignment angles. Both numerical simulations and experiments are conducted to validate the proposed method.