The article discusses a new filter design known as the multi state constraint equivariant filter (MSC EqF) for visual-inertial navigation systems (VINS). This filter design addresses the problem of visual-inertial odometry (VIO) including IMU bias, and camera intrinsic and extrinsic calibration states. It improves upon existing techniques by incorporating all states in the group structure, leading to improved behavior, particularly when linearization points deviate significantly from the truth. The approach also eliminates the need for error uncertainty adaptation, observability constraint, or other consistency enforcing techniques. The filter’s performance is evaluated using three different real-world datasets.
Publication date: 21 Nov 2023
Project Page: Not provided
Paper: https://arxiv.org/pdf/2311.11649
