Robotic Vision Summer School 2018

Tutorial on Object SLAM

The goal of Simultaneous Localization and Mapping (SLAM) is to construct the representation of an environment while localizing the robot with respect to it. However, this does not provide any understanding of the physical world that the robot is moving in. Such understanding is important for meaning interactions with the world and also results in improved performance for the mapping and localization tasks. This tutorial will introduce the problem of SLAM at it current stage of development and then address various development towards a semantic understanding of the world and its effects on the original SLAM formulation.

Program

The tutorial is organized as four 45 min sessions that start from the basics of SLAM and progressively build toward the goal of Semantic SLAM.

Speaker Topic
Viorela Ila SLAM in a nutshell Slides
Yasir Latif Tracking and Data Association Slides
Trung Pham Semantic Mapping Slides
Vincent Lui Semantic SLAM Slides
Speakers

Viorela Ila

I am a research fellow with the ARC Centre of Excellence for Robotic Vision at the Australian National University (ANU). My research interests span from robot vision to advanced techniques for simultaneous localization and mapping (SLAM) and 3D reconstruction based on cutting-edge computational tools such as graphical models, modern optimization methods and information theory.

I received the Engineering degree in Industrial Engineering and Automation from the Technical University of Cluj-Napoca, Romania, in 2000 and the Ph.D. in Information Technologies from the University of Girona, Spain, in 2005. After the PhD studies, I joined the Robotics group at the Institut de Robótica i Informàtica Industrial, Barcelona, Spain. In 2009 I have been awarded the MICINN/FULBRIGHT post-doctoral fellowship which allowed me to join the group of Prof. Frank Dellaert at College of Computing, Georgia Tech, Atlanta US. In 2010, I joined the robotic group at LAAS-CNRS, Toulouse, France to work in the ROSACE project founded by RTRA-STAE. Between 2012 and 2014 I was research scientist at Brno University of Technology in Czech Republic.

Yasir Latif

Yasir Latif did his bachelors at Ghulam Ishaq Khan Institute of Engineering Science and Technology in Topi, Pakistan and his master in Communication Engineering from Technical University of Munich (TUM), Germany. After that, he pursued his PhD at University of Zaraogoza, Spain under the supervision of Prof. Jose Neira. He visited Imperial College London and Massachusetts Institute of Technology for short research stays during that period. The main theme of his doctoral thesis was reliable loop closure detection and verification for the Simultaneous Localization and Mapping (SLAM) problem. His interests include SLAM, Computer Vision and looking for the ultimate question.

Trung Pham

Trung Pham is a research fellow at the Australian Centre of Robotic Vision. He earned his Ph.D. degree in Computer Vision in 2014 from the University of Adelaide, Australia. He received a Google PhD Fellowship in 2012 for his excellent PhD research. In 2013, he was one of 200 young researchers worldwide invited to attend the first-ever Heidelberg Laureate Forum. Trung’s main research interests are in the fields of computer vision, machine learning and robotic vision. His research outcomes have been published in highest-ranking journals and conferences across computer vision, machine learning and robotics such as TPAMI, CVPR, ICCV, NIPS, ICRA.

Vincent Liu

Vincent is currently a research fellow at Monash University. He did his PhD under the supervision of Professor Tom Drummond at the same university, where he investigated different ways of performing pose estimation and developed efficient and robust techniques for pose estimation in monocular visual odometry and SLAM problems. His research interests lie in the area of SLAM, 3D reconstruction, pose estimation, and augmented reality.

videos

State of the art SLAM systems

ORB-SLAM2 DSO LSD-SLAM KinectFusion ElasticFusion

Semantic SLAM systems

SemanticFusion Dense SLAM Segmentation