Object detection and tracking aided SLAM in image sequences for dynamic environment.

Object detection in a dynamic environment is important for accurate tracking and mapping in Simultaneous Localization and Mapping (SLAM). Dynamic feature points from people or vehicles are the main cause of unreliable SLAM performance. Previous researchers have used varied techniques to solve this problem, such as semantic segmentation, optical flow, and moving consistency check algorithm. In this proposal, Object Detection and Tracking SLAM (ODTS), we define a weighted grid-based attention model for a feature tracking module to track landmarks and objects. ODTS system tracks landmarks, such as buildings in the background, and objects, such as vehicles, in the foreground. For optimizing performance, a robust self-attention module is integrated. For evaluation, the trajectory of the robot is tracked, and the root mean square error (RMSE) is recorded. Additionally, the number of background and foreground feature points were observed for landmarks and objects. ODTS significantly minimizes the tracking lost problem and produces more accurate maps and tracking of feature points.