Environment Modeling for Cooperative Aerial/Ground Robotic Systems

摘要: This paper addresses the cooperative localization and visual mapping problem for multiple aerial ground robots.We propose use of heterogeneous landmarks, points line segments. A large-scale SLAM algorithm is generalized to manage robots, in which a global graph maintains topological relationships between series local sub-maps built by different robots. Only single camera setups are considered: order achieve undelayed initialization, we present novel parametrization lines based on anchored Plucker coordinates, add extensible endpoints enhance their representativeness. The maps combine such with 3D parametrized inverse-depth. overall approach evaluated real-data taken helicopter rover an abandoned village.