TY  - CHAP
ID  - davfercha_IWANN_2019_integration
T1  - Integration of CNN into a Robotic Architecture to Build Semantic Maps of Indoor Environments
A1  - Fernandez-Chaves, David
A1  - Ruiz-Sarmiento, J. R.
A1  - Petkov, Nicolai
A1  - Gonzalez-Jimenez, Javier
T3  - Advances in Computational Intelligence. IWANN 2019. Lecture Notes in Computer Science
Y1  - 2019
VL  - 11507
SP  - 313
EP  - 324
PB  - Springer, Cham
SN  - 978-3-030-20517-1
UR  - https://drive.google.com/uc?id=1ed-sxK8OPM7fG1rwqp_y0GGKxBC8DktF&export=download&authuser=0
M2  - doi: 10.1007/978-3-030-20518-8_27
N2  - In robotics, semantic mapping refers to the construction ofa rich representation of the environment that includes high level infor-mation needed by the robot to accomplish its tasks. Building a semanticmap requires algorithms to process sensor data at different levels: geo-metric, topological and object detections/categories, which must be inte-grated into an unified model. This paper describes a robotic architecturethat successfully builds such semantic maps for indoor environments. Forthis purpose, within a ROS-based ecosystem, we apply a state-of-the-artConvolutional Neural Network (CNN), concretely YOLOv3, for detect-ing objects in images. The detection results are placed within a geometricmap of the environment making use of a number of modules of the ar-chitecture: robot localization, camera extrinsic calibration, data form adepth camera, etc. We demonstrate the suitability of the proposed frame-work by building semantic maps of several home environments from theRobot@Home dataset, using Unity 3D as a tool to visualize the maps aswell as to provide future robotic developments.
M1  - img_url=http%3A%2F%2Fdrive.google.com%2Fuc%3Fexport%3Dview%26id%3D1z-fgNZjndGTWjWgsNlI71X-CnNz_a8bD
M1  - rank_indexname=
M1  - rank_pos_in_category=
M1  - rank_num_in_category=
M1  - rank_cat_name=
M1  - impact_factor=
ER  -