SPINS: A structure priors aided inertial navigation system

We propose a navigation system combining sensor-aided inertial and prior-map-based localization to improve the stability accuracy of robot in structure-rich environments. Specifically, we adopt point, line, plane features enhance feature richness low-texture environments reliability. additionally integrate structure prior information constrain drifts accuracy. The is called priors parameterized as low-dimensional relative distances/angles between different geometric primitives. formulated graph-based optimization problem that contains sliding-window-based variables factors, including Inertial Measurement Unit, heterogeneous features, priors. A limited number are selected based on gain alleviate computation burden. Finally, proposed framework extensively tested synthetic data, public data sets, and, more importantly, real Unmanned Aerial Vehicle flight obtained from both indoor outdoor inspection tasks. results show scheme can effectively robustness for autonomous robots civilian applications.