3D Modeling with a Moving Tilting Laser Sensor for Indoor Environments

Abdennour Aouina (1) , Michel Devy (1) , A Marin Hernandez (2)

1 LAAS-RAP - Équipe Robotique, Action et Perception
2 Universidad Veracruzana

Many works are devoted to 3D modeling of indoor environments from mobile sensors. This function has been performed using multiple robotic platforms, equipped with different types of 3D sensors; many theoretical approaches have been proposed to refine these models, either based on SLAM algorithms when considering only sparse features, or based of ICP-based methods when a dense model is made from the registration of raw 3D data. This paper presents an approach to build a 3D surfacic model based on planar surfaces, using a tilting LRF (Laser Range Finder) mounted on the PR2 mobile robot, by the fusion of ribbons (sequence of aligned surfels) extracted from the successive scan lines. These lines are acquired on the fly from the LRF, avoiding a stop and go strategy. The ribbons are aggregated using the robot positions given by a 2D SLAM process executed independantly and simultaneously; all required information are memorized so that the surfacic model can be corrected when the SLAM process corrects the robot trajectory after a loop closure.

Domaines: Informatique [cs]; Vision par ordinateur et reconnaissance de formes [cs.CV]; Robotique [cs.RO]

Abdennour Aouina, Michel Devy, A Marin Hernandez. 3D Modeling with a Moving Tilting Laser Sensor for Indoor Environments. IFAC World Congress 2014, Aug 2014, Capetown, South Africa. ⟨10.3182/20140824-6-ZA-1003.00460⟩. ⟨hal-02049127⟩