Camera Motion Estimation from RGB-D-Inertial Scene Flow

Samuel Cerezo, Javier Civera; Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2024, pp. 841-849

In this paper we introduce a novel formulation for cam- era motion estimation that integrates RGB-D images and in- ertial data through scene flow. Our goal is to accurately es- timate the camera motion in a rigid 3D environment along with the state of the inertial measurement unit (IMU). Our proposed method offers the flexibility to operate as a multi- frame optimization or to marginalize older data thus ef- fectively utilizing past measurements. To assess the perfor- mance of our method we conducted evaluations using both synthetic data from the ICL-NUIM dataset and real data sequences from the OpenLORIS-Scene dataset. Our results show that the fusion of these two sensors enhances the accu- racy of camera motion estimation when compared to using only visual data.