Limits...
Towards the automatic scanning of indoors with robots.

Adán A, Quintana B, Vázquez AS, Olivares A, Parra E, Prieto S - Sensors (Basel) (2015)

Bottom Line: The document presents the principal steps of the process, the experimental setup and the results achieved.This system has been tested under real conditions indoors with promising results.The future is addressed to extend the method in much more complex and larger scenarios.

View Article: PubMed Central - PubMed

Affiliation: Visual Computing and Robotics Lab, Universidad de Castilla-La Mancha (UCLM), Paseo de la Universidad, 4, Ciudad Real 13071, Spain. Antonio.Adan@uclm.es.

ABSTRACT
This paper is framed in both 3D digitization and 3D data intelligent processing research fields. Our objective is focused on developing a set of techniques for the automatic creation of simple three-dimensional indoor models with mobile robots. The document presents the principal steps of the process, the experimental setup and the results achieved. We distinguish between the stages concerning intelligent data acquisition and 3D data processing. This paper is focused on the first stage. We show how the mobile robot, which carries a 3D scanner, is able to, on the one hand, make decisions about the next best scanner position and, on the other hand, navigate autonomously in the scene with the help of the data collected from earlier scans. After this stage, millions of 3D data are converted into a simplified 3D indoor model. The robot imposes a stopping criterion when the whole point cloud covers the essential parts of the scene. This system has been tested under real conditions indoors with promising results. The future is addressed to extend the method in much more complex and larger scenarios.

No MeSH data available.


(a) Labeled map for Cycles 1 and 2 after applying the NBV algorithm; (b) labeled voxel space for Cycles 1 and 2.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4481921&req=5

f7-sensors-15-11551: (a) Labeled map for Cycles 1 and 2 after applying the NBV algorithm; (b) labeled voxel space for Cycles 1 and 2.

Mentions: Figure 7a shows the results achieved after applying the NBV algorithm. The new position of the scanner is marked in the earlier labeled map, and the updated labeled map is shown on the right. Note that some red pixels (occluded) have turned into white pixels (empty), and consequently, the entropy of the scene has increased. Figure 7b shows the three-dimensional voxel space in which some regions (pointed out in the figure) are relabeled as occupied voxels.


Towards the automatic scanning of indoors with robots.

Adán A, Quintana B, Vázquez AS, Olivares A, Parra E, Prieto S - Sensors (Basel) (2015)

(a) Labeled map for Cycles 1 and 2 after applying the NBV algorithm; (b) labeled voxel space for Cycles 1 and 2.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4481921&req=5

f7-sensors-15-11551: (a) Labeled map for Cycles 1 and 2 after applying the NBV algorithm; (b) labeled voxel space for Cycles 1 and 2.
Mentions: Figure 7a shows the results achieved after applying the NBV algorithm. The new position of the scanner is marked in the earlier labeled map, and the updated labeled map is shown on the right. Note that some red pixels (occluded) have turned into white pixels (empty), and consequently, the entropy of the scene has increased. Figure 7b shows the three-dimensional voxel space in which some regions (pointed out in the figure) are relabeled as occupied voxels.

Bottom Line: The document presents the principal steps of the process, the experimental setup and the results achieved.This system has been tested under real conditions indoors with promising results.The future is addressed to extend the method in much more complex and larger scenarios.

View Article: PubMed Central - PubMed

Affiliation: Visual Computing and Robotics Lab, Universidad de Castilla-La Mancha (UCLM), Paseo de la Universidad, 4, Ciudad Real 13071, Spain. Antonio.Adan@uclm.es.

ABSTRACT
This paper is framed in both 3D digitization and 3D data intelligent processing research fields. Our objective is focused on developing a set of techniques for the automatic creation of simple three-dimensional indoor models with mobile robots. The document presents the principal steps of the process, the experimental setup and the results achieved. We distinguish between the stages concerning intelligent data acquisition and 3D data processing. This paper is focused on the first stage. We show how the mobile robot, which carries a 3D scanner, is able to, on the one hand, make decisions about the next best scanner position and, on the other hand, navigate autonomously in the scene with the help of the data collected from earlier scans. After this stage, millions of 3D data are converted into a simplified 3D indoor model. The robot imposes a stopping criterion when the whole point cloud covers the essential parts of the scene. This system has been tested under real conditions indoors with promising results. The future is addressed to extend the method in much more complex and larger scenarios.

No MeSH data available.