Learning to Segment Dynamic Objects using SLAM Outliers
This addresses the challenge of SLAM robustness in dynamic environments for robotics and autonomous systems, though it is incremental as it builds on existing SLAM methods with a novel segmentation approach.
The paper tackles the problem of dynamic objects interfering with SLAM by learning to segment them using SLAM outliers, requiring only one monocular sequence per object for training, and integrates this into SLAM systems to remove features on dynamic objects, achieving better performance than State-of-the-Art on the TUM RGB-D dataset in monocular mode and a new stereo dataset.
We present a method to automatically learn to segment dynamic objects using SLAM outliers. It requires only one monocular sequence per dynamic object for training and consists in localizing dynamic objects using SLAM outliers, creating their masks, and using these masks to train a semantic segmentation network. We integrate the trained network in ORB-SLAM 2 and LDSO. At runtime we remove features on dynamic objects, making the SLAM unaffected by them. We also propose a new stereo dataset and new metrics to evaluate SLAM robustness. Our dataset includes consensus inversions, i.e., situations where the SLAM uses more features on dynamic objects that on the static background. Consensus inversions are challenging for SLAM as they may cause major SLAM failures. Our approach performs better than the State-of-the-Art on the TUM RGB-D dataset in monocular mode and on our dataset in both monocular and stereo modes.