eversvd,+4 (15)

This study proposes the use of heterogeneous visual landmarks, specifically points and line segments, to achieve effective cooperation in indoor SLAM environments. The primary objective is to address the challenge of un-delayed initialization required by bearing-only observations. To tackle this, the well-known inverse-depth parameterization is adopted for the estimation of 3D points. Additionally, a novel parameterization based on anchored Plücker coordinates is introduced for estimating 3D line segments, incorporating extensible endpoints. By utilizing both point and line features, the proposed vision-based SLAM system is designed to operate efficiently within indoor environments, overcoming limitations associated with traditional point feature-based systems that may struggle with dynamic obstacles. As such, the system demonstrates the potential to perform SLAM even with significantly fewer features compared to point-only approaches. The findings highlight the practicality and effectiveness of leveraging heterogeneous features in mobile robotics, contributing to advancements in achieving greater autonomy for robots navigating complex indoor spaces.