José Lamarca
|
PhD. Student in Computer Vision, Machine learning and 3D reconstruction at the Computer Science and Systems Engineering Department of the University of Zaragoza, Spain. Email: jlamarca@unizar.es Tel: (+34) 691 70 9996 |
 |
|
I am currently PhD. Student in Computer Science and System Engenering at the University of Zaragoza advised by Prof. Jose Maria M. Montiel My research interests now are: Non-Rigid Structure from Motion, SLAM, Shape-from-template, Visual SLAM, Robotics, deep learning... I finished my Bs. in Industrial Technologies Engineering at July 2014 in Zaragoza and my M.Sc. in Engineering Specialised in Industrial Automation and Robotics in 2016. I got the opportunity of a internship in KUKA Roboter GmbH. There I developed my Master Thesis advisored by Dr. Javier Civera Monocular SLAM and Active vision in mobile platforms. I defended my Master thesis in December 2016. At the end of my Master a I got a position as PhD candidate in the University of Zaragoza to start a researching about the extension of visual SLAM to deformable environments. For a more detailed information, you can download my CV. |
|
|
abstract |
bibtex
The current SLAM algorithms cannot work without assuming rigidity. We propose the first real-time tracking thread for monocular VSLAM systems that manages deformable scenes. It is based on top of the Shape-from-Template (SfT) methods to code the scene deformation model. Our proposal is a sequential method that manages efficiently large templates, i.e. deformable maps estimating at the same time the camera pose and it can be relocalised in case of tracking loss. We have created a new dataset to evaluate our system. Our results show the robustness of the method in deformable environments while running in real time with errors under 3\% in depth estimation. |
|
abstract |
bibtex
This paper describes the structure of a mobile robot suitable for laboratory experiments. The robot is developed with the ideas of minimizing the production costs and of executing motion commands received from an external computer that runs the (possible computational intensive) planning algorithms. The proposed system is a differential wheeled robot built based on an Arduino platform, it has a small physical size and it is capable of wireless communication. These characteristics make it useful for validating planning algorithms for mobile robots in computer-guided laboratory experiments, while lowering the expenses required by such a setup. Moreover, in order to compensate the communication delay between the central computer and the robot, an estimation algorithm is implemented on the robot for obtaining a smoother motion. |
Template: pathak, barron, akar, sgupta, donahue and j.m. Facil |