Direct visual-inertial odometry with semi-dense mapping

The paper presents a direct visual-inertial odometry system. In particular, a tightly coupled nonlinear optimization based method is proposed by integrating the recent development in direct dense visual tracking of camera and the inertial measurement unit (IMU) pre-integration. Then a factor graph optimization is adopted to estimate the pose and position of the camera, and a semi-dense map is created simultaneously. Two sliding windows are maintained in the proposed approach. The first one, based on direct sparse odometry (DSO), is to estimate the depths of candidate points for mapping and dense visual tracking. In the second one, measurements of both the IMU pre-integration and direct dense visual tracking are fused probabilistically based on a tightly-coupled, optimization-based sensor fusion framework. As a result, the scale drift of visual odometry is compensated by the constraints from the IMU pre-integration. Evaluations on real-world benchmark datasets show that the proposed method achieves competitive results in indoor scenes.