Daniel Canelhas

Daniel Canelhas

I am a PhD student in robot vision, interested in 3D mapping, camera tracking, optimization and machine learning. I like to make art, write code, play music, do sports, read science-fiction, science-fact, play video-games, and tell bad jokes. 

Academic

  • 2012–present
    Ph.D. in Computer Science at Örebro Universitet
    Truncated Signed Distance Fields and Their Application to Robotics
  • 2010–2012
    M.Sc. Robotics and Intelligent Systems at Örebro Universitet
    Scene Representation, Registration and Object Detection in a Truncated Signed Distance Function Representation of 3D Space 
  • 2002–2006
    B.Sc. Mechanical Engineering with emphasis in Mechatronics at Pontíficia Universidade Católica – Minas Gerais – Brazil
    Undergraduate research with Prof. Marcelo Becker on auto-localization of indoor mobile robots under a scientific initiation scholarship

Professional

  • 2016 – Örebro Universitet, Teaching Assistant in Computer Graphics
  • 2016 – Örebro Universitet, Teaching Assistant in Multi-Agent Systems
  • 2015 – Örebro Universitet, Undergraduate Thesis supervisor
    “Baking And Compression For Dynamic Lighting Data” by Tom Olsson – DIVA 
  • 2015 – Örebro Universitet, Teaching Assistant in Computer Animation
  • 2014 – Örebro Universitet, Teaching Assistant in Computer Graphics
  • 2013 – Örebro Universitet, Teaching Assistant in C-programming
  • 2007–2012 – Areva T&D / Alstom Grid (Västerås), Mechanical Design Engineer in High-voltage AC substation (15 to 400kV)
  • 2006 – CEMIG (Belo Horizonte, Brazil), Internship in High-voltage AC substation automation and electrical engineering
  • 2001–2002 – Intervalo (Belo Horizonte, Brazil), Internship in classical 2D animation and illustration

SDF tracker : a parallel algorithm for on-line pose estimation and scene reconstruction from depth images. Or, alternatively, my Master's thesis

3D reconstruction using the SDF-Tracker algorithm, estimating camera pose and fusing depth images into a dense 3D model on the CPU (code)

An Eigenshapes Approach to Compressed Signed Distance Fields and Their Utility in Robot Mapping

Mapping a large environment by shifting the reconstruction volume and encoding exiting voxels as their eigen-shapes descriptors.

From Feature Detection in Truncated Signed Distance Fields to Sparse Stable Scene Graphs.

For really large scenes you might want an (even) sparser representation. Sparse Stable Scene Graphs (SSSG) represents the dense 3d model as a graph of spatially interconnected feature

 

 

Contact

Dr. Daniel Canelhas

AASS Research Centre
School of Science and Technology
Örebro University
70182 Örebro, Sweden
Room T1102
Phone +46 (0)19 30 35 36
daniel.canelhas@oru.se

Publications

Journal Articles