Observabilty-Aware Trajectories for Geo...

In this paper, we apply an observability-aware trajectory generation method to the estimation of geometric and inertial parameters of an Unmanned Aerial System (UAS). These parameters are critical for reliable control and agile maneuvers, especially in the context of reconfiguration of the aerial vehicles during manipulation or transportation tasks. An extended observability      analysis provides detailed insights on the observable and inter-state dependencies. We employ the observability-aware motion generation approach considering full system dynamics and self-calibration parameters. Improvements in the absolute error of r_BP estimates of up to 46.8% and decreases in uncertainty of up to 87% are achievable with this approach. Experiments with an autonomous quadrotor platform validate the approach.