Comparison and Analysis of Feature Method and Direct Method in Visual SLAM Technology for Social Robots

Abstract

Simultaneous localization and mapping (SLAM) technology is adopted for localization and environmental map reconstruction in unknown environments. Visual SLAM provides rich three-dimensional map information that helps social robots complete complex tasks. Visual odometry (VO) is an important component of Visual SLAM. Here, we first review the principles of the feature and direct methods of VO. Thereafter, we introduce a social robot platform for testing the two VO methods. We detail the implementation and setup of the feature and direct methods of VO. Finally, we compare and analyze the two methods and provide suggestions for their applications.